Patron of the Mind: Jeffrey Epstein, Intelligence and Consciousness, 2009–2019

What his science funding points to, who pushed back, and who took part

A research paper built from the Jmail archive and checked against the U.S. Department of Justice originals. Version final-r3, prepared 11 October 2026 (CT).

Abstract

Between 2009 and 2019 Jeffrey Epstein, a financier who in 2008 pleaded guilty to Florida state charges of soliciting prostitution and of procuring a person under 18 for prostitution [S98], wrote about 1,500 distinct passages on intelligence and the mind: machine intelligence, the brain, genes, quantum physics and consciousness. This paper scanned all 1,755,621 non-promotional emails in the Jmail archive, together with its document, message, purchase and photo collections. It separates what Epstein himself wrote from what others wrote to him, and every quoted email was checked against the government's own PDF (128 government originals read in all; 117 cited).

The record does not show a man building artificial intelligence. It shows a patron organised around one recurring idea: that living things, brains included, communicate in hidden, encrypted signals, and that deception is how life protects itself. He wanted to pay code-breakers, including people from the National Security Agency, to "intercept communicaiton between living cells". Machine AI was a small, cheap and early part of the portfolio. The larger flows went to evolutionary mathematics at Harvard, the MIT Media Lab, genome projects with George Church, and, late on, quantum theories of consciousness. At the far edge, in 2018, he discussed investing in a "designer baby and human cloning company", writing that "i have no issue with investing the problem is only if i am seen to lead". No payment for it has been found. Separately, a summary chart of Deutsche Bank's records, e-mailed in August 2019 by Justin Alfano to recipients whose names are redacted ([R-12]), lists a 2014 payment of $500,001 to "Joichi Ito" from an entity in Epstein's account charts, with wire details reading "subcr to Kyara Investments III, LLC to fund Blockstream Investment". The chart does not show who controlled that fund, or whether or when the money reached Blockstream.

Three sides are shown throughout: his aims in his own words; the pushback he met and the decisions he made anyway; and the scientists and institutions who took part, with what they said then and later. Pushback at the time came from institutions and outsiders (Harvard, which refused his gifts after his conviction; a prosecutor; a reputation firm; TED; Craig Venter's office) and, among the scientists, as technical disagreement, most sharply from Noam Chomsky. Moral refusals came mostly after 2019. The pattern the funding points to is a mix of real curiosity, purchased standing, and a steady preference for giving in ways that kept his name off the gift.

Contents

Name & Redaction Legend · Reader's note · 1. Introduction · 2. Data and methods · 3. The portfolio · 4. Biology as code and deception · 5. Machine minds · 6. Quantum minds · 7. The genome · 8. Designer babies and the far edge · 9. Money, anonymity and control · 10. Science as rehabilitation · 11. Then and later · 12. Explanations, limits and conclusion · Appendix A: Jmail-only text · Appendix B: Verification table · Appendix C: Table J1 · Appendix D: Money flows · Appendix E: Glossary · References

Name & Redaction Legend

This legend explains every label used in place of a name. Names that the government originals print unredacted are given in full (rule 1). A black box in a government original is shown as a numbered redaction, for example [R-5], and nothing under a box is ever guessed (rule 3). No cited passage names or describes a possible victim or minor, so no Jane or John Doe label was needed (rule 2). The screen is described in the last row. Where Jmail shows a name that the original redacts, both are shown in Table J1 (rule 4). Each label in the text links to its row here.

Names visible in the originals and used in full include Judith Donath, Sebastian Seung, Theresa Tomic, Bryan Bishop, Austin Hill, Gregory Markel, Wendy Woska, Lisa M. Betting (in Craig Venter's office), Marissa Wozniak, Henry Jarecki, Lesley Groff, Richard Kahn, Tom Pritzker, Maurice Sonnenberg, Justin Alfano, Andrew Stemmer, Erika Engelson and Vijay Dewan (the last three are copied on the Deutsche Bank chart e-mail), and David Ingram (a Reuters reporter). Neri Oxman appears in the original as "Neri"; her full name is from MIT's public report [S04].

Reader's note: labels, links and how terms are explained

Every claim about an email carries its document number, a link to the government PDF on justice.gov, and a trust label:

• OC (original confirmed): the quoted words, the author and the order of messages were read in the justice.gov PDF.
• TP (text present): the words are in the justice.gov PDF, but the PDF shows no sender, recipient or date and does not say who wrote them (for example an unsigned draft).
• CONFLICT: the Jmail version and the government original differ in a way that matters. Both versions are shown in Table J1.
• ST (site text): text that exists only in Jmail, such as the iMessage collection, which has no government original. ST quotes appear only in Appendix A, and no chapter or section title uses them.
• SEC: an outside source (news report, institutional report, statement or scientific paper), cited as [Sxx] and listed in the References.

Technical terms are explained in the sentence where they first appear, and briefly again if they come back much later. A glossary at the back repeats those explanations for reference only. Epstein's own words are quoted exactly as he typed them, misspellings included, followed by a plain reading of what he seems to mean.

1. Introduction: a patron of the sciences of the mind

1.1 The puzzle

In January 2015 David Ingram, a Reuters reporter preparing a story about Jeffrey Epstein, put questions about Epstein's philanthropy to his lawyer, Martin G. Weinberg, who forwarded them to Epstein. Epstein drafted wording, in the third person, for how that philanthropy might be described: "quantum computing, artificial intelligence, evolutionary dynamics, membrane computing, mathematical biology, synthetic general intelligence, mathematics of cooperation. genetics, encryption algorithms , might better portray his intellectual philanthropic interests" (EFTA00699570, OC). Several items need a word of explanation. Evolutionary dynamics is the mathematics of how strategies and genes spread through populations over time. Membrane computing is a branch of computer science that models computation on the way living cells pass substances across their membranes. Mathematical biology uses equations to describe living systems. He was by then a registered sex offender. Six years earlier he had served a jail sentence in Florida after pleading guilty to soliciting prostitution and to procuring a person under 18 for prostitution [S98]. Yet the list he chose for the press was a list of frontier sciences of the mind and of life. This paper is about that science patronage. It is not about his abuse of girls and young women, which courts, prosecutors and survivors have documented elsewhere. But it never forgets that context: the same years, money and social standing are involved.

This paper asks what that choice, and the money behind it, actually points to. Did Epstein have a project in artificial intelligence, the field that tries to build machines that think? Was there a programme behind his talk of "superintelligence", meaning an intellect far beyond the best human minds, and of the "ubermensch"? Or was science mainly a way to buy standing and company? The question matters for the history of science funding as well as for Epstein's story: many of the people he paid were, and are, leading researchers.

1.2 What the archive lets us see

The Jmail archive is a searchable public copy of the Epstein files released by the U.S. Department of Justice and the House Oversight Committee [S78, S87]. Its emails reach from the early 2000s to 2019, but they become dense only from 2009. It also holds documents, a small collection of iMessages, a record of Amazon purchases and photo descriptions. The archive is huge and uneven. It shows what Epstein typed and what people typed back. It does not show phone calls or meetings, and it rarely shows money actually changing hands. Bank transfers appear only where a document mentions them.

1.3 Three sides

This paper sets out to show every side. Each theme chapter therefore ends with three parts:

1. His ideals and aims, in his own words. What he said he wanted to fund and why.
2. Pushback, and the decisions he made anyway. Who declined, warned, rebutted or flagged him, and what he did next.
3. Supporters and beneficiaries, then and later. Who took part or took money, how they described it at the time, and what they have said since 2019.

1.4 The argument in brief

The archive does not show an AI builder. It shows a patron whose most persistent idea was that life is code, and code is defended by deception. Cells, brains and societies, in his telling, send encrypted signals, and the way to understand them is to break the codes, as an intelligence agency would. This idea links almost everything he funded: evolutionary game theory at Harvard (the mathematics of how competing strategies fare as populations evolve), the Media Lab at MIT, artificial general intelligence (AGI, a machine with flexible, human-level thinking) in Hong Kong, the genome work of George Church, and quantum theories of consciousness. Around it sits a second, practical pattern: he preferred to give in ways that hid his name, and he used science to repair his reputation after his conviction. At the far edge sits a third: talk of remaking humans, including a 2018 discussion of a "designer baby" company in which, he wrote, "the problem is only if i am seen to lead".

Chapter 2 sets out the data and methods. Chapters 3 to 8 take the themes in turn: the portfolio as he described it; biology as code and deception; machine minds; quantum minds; the genome; and designer babies and human enhancement. Chapter 9 follows the money and the anonymity. Chapter 10 shows science used for legal and reputational rehabilitation. Chapter 11 puts the participants' then-and-later statements side by side. Chapter 12 weighs four competing explanations and concludes.

2. Data, methods and rules of evidence

2.1 Sources and coverage

Table 1 lists what was scanned. The email scan covered every non-promotional message. The other apps were covered through their bulk datasets where these existed. Where they did not, server-rendered pages were fetched politely, with ten seconds between requests and no use of the site's private interfaces.

Table 1. Coverage by app

2.2 Finding the topic

The search began with a seed list of terms (AI, consciousness, genome, quantum, and so on). It then grew by co-occurrence: words that turned up alongside the seed terms more often than chance were reviewed and, if relevant, added. Thirteen further term families came in this way, including deception, signal intelligence, power laws, membrane computing and music-and-the-brain. The final vocabulary produced 36,775 raw hits. After removing exact and near-duplicates, idioms ("quantum leap"), news digests and signature blocks, there were 2,084 distinct statements written by Epstein and 2,945 distinct statements by correspondents in threads he took part in, spread over 7,041 threads.

2.3 Who wrote what

Email threads quote earlier messages, so the same words can appear under many senders. A script cut each message at the first quoted-reply marker, so that only the newly written text was attributed to the sender. Every quotation in this paper was then re-read in the government PDF to confirm who wrote it and in what order (§2.5).

2.4 How precise the counts are

Counts from keyword searches include false hits. To measure this, samples were hand-coded. For the topic families, 140 rows (10 per family) were coded, and 72% were relevant overall. Precision ranged from 100% (evolution and cooperation) down to 50% (consciousness, and physics). That implies roughly 1,500 relevant distinct statements by Epstein. The largest families were mathematics (about 422), the brain (325), evolution (227), intelligence (222), biodata (187), computation (157) and AI (127). Quantum (78), energy (37) and longevity (18) were smaller.

For the three sides, a second sample of 60 rows was drawn (20 per side, at random) from the 'pushback', 'ideals' and 'supporters' keyword searches, and hand-coded. Precision was very low: 0 of 20 for pushback (95% interval 0–16%), 2 of 20 for ideals (3–30%) and 6 of 20 for supporters (15–52%). Words like "declined", "reputation" or "grant" turn up mostly in news digests, legal mail and unrelated fundraising. The raw side counts are therefore not used as evidence in this paper. Every triad item below was found by the searches but then read, selected and checked by hand. The curated set has 86 core items plus those added in this phase (data/triad_evidence.csv, data/precision_sides_coded.csv).

2.5 Verification

Every email quoted or relied on was fetched from justice.gov and read in its PDF, with requests spaced five to ten seconds apart. In all, 128 government originals were checked in six rounds (two in the first phase, one in this phase, one during the internal review, one during the first outside review, which fetched four more originals, among them the three money-table documents that had previously been read only in Jmail, and one during the second outside review, which fetched ten more: the December 2013 to March 2014 Gershenfeld "deception" thread and the January 2015 Kahn messages about Goertzel). For documents whose text layer was unclear or that involved redactions, the page image was viewed. Four items first seen only in the House Oversight release were matched to Department of Justice duplicates of the same text, and those duplicates were verified instead. At the end, every quotation in the main text was checked by machine against the PDF text of the documents cited beside it, and the author line and message order were then read by hand (Appendix B). Of 363 quoted strings, 287 matched the PDF text exactly and 35 matched apart from optical-reading noise (a fuzzy score of 90 or more out of 100). The remaining 41 are quotations from outside sources, titles or names, which were checked against those sources. A last pass rendered the page holding each quoted email string as an image and read it by optical character recognition: 313 of 322 strings matched the image text, and the other 9, all on two pages whose typewriter-style text the recognition software could not read (EFTA00770109 and EFTA00372888), were confirmed by eye, along with their order in the thread. For every quotation that is new or changed in this revision, the author line and the order of messages were read on the page image. That check changed the order of the July 2018 Bishop–Hill exchange (§8.3), added the first appearance of the race claim in Bach's own 10 July 2016 message (§5.5, §8.1), and corrected one word ("liikes", §3.5). An earlier pass had corrected the order of the February 2013 Templeton exchange (§3.5) and a phrase in Chapter 12.

2.6 What Jmail's own labels were used for

Jmail adds machine-written summaries, topic tags, crowd "stars" and inferred names. These were used only as leads. Where Jmail's display differs from the government original, both are shown in Table J1 and neither is treated as the more authoritative.

2.7 The shape of the record

Figures 1 and 2 show almost nothing before 2009, a step up in 2009, a plateau from 2010 to 2014, and a peak from 2015 to 2018. The low early numbers partly reflect which emails survive, so they are not a measure of interest. Within the record, machine AI comes early and small. The quantum, consciousness and genome strands come late, from 2014 onwards.

3. The portfolio: how he described what he was buying

Argument. From 2009 Epstein kept re-listing the same bundle of fields, both to plan funding and to describe himself. Machine AI is one small item on those lists. "Biology and deception" is the recurring headline. The lists doubled as his public identity.

3.1 The 2009 "gatherings" list

On 11 August 2009 Epstein wrote to the literary agent John Brockman, whose Edge Foundation gathers well-known scientists for discussion: "I intend to fully fund the following gatherings ( meetings with lots of interaction hikes coffee. cross pollination ) over the next year". First on his list was "Evolutionary Code dynamics.. a brand new field that encompasses, game theory, virus dynamics, (both computers and biological). , cryptography , and genetics". Game theory here means the mathematics of strategy, how competing players should act. Second was "Synthetic General Intelligence.. encompassing, signal processing, statistical and machine learning, virtual world robots" (EFTA02441292, OC). Signal processing is the engineering of pulling information out of signals such as sound, radio or sensor readings. Even at the start, the AI item was framed through signals and codes.

A month later he asked the New York Times technology reporter John Markoff to help him "find a good writer to put together a summary of my science projects, past and current„ quantum computing, string theory , loop quntum gravity, chinese chi , complexity, neuroscience, bio physics, evolutionary dynamics, cosmology, cognitve neuro, foundations of morals" (EFTA02437570, OC). String theory and loop quantum gravity are rival attempts to unite physics' theory of gravity with quantum mechanics. "Chinese chi", the life-energy of traditional Chinese medicine, sits in the same list without distinction. In January 2010 he told Barnaby Marsh, then an officer of the John Templeton Foundation (a large charity that funds research at the meeting point of science and religion), that "my funding is now concentrated on 1 decoding and evolution.. ( there are no energy equations in predator prey models. ),2 . music and the brain" and psychoactive drugs "used as tools to probe the mind, not the brain" (EFTA02433017, OC).

3.2 Writing himself as a science patron

The lists were also about how he would be seen. In October 2010 his associate Al Seckel, a writer on optical illusions who organised conferences for him, turned "a stream of consciousness about your funding interests for 2011" into a public-facing list. Epstein asked him to "add new financial systems, important.. add pre life as replication of signals" (EFTA00753978, OC). The next chapter returns to that idea of "signals". Chapter 10 returns to what Seckel was doing with the list.

3.3 The 2015–2016 lists

In June 2015 Epstein told the biologist Rupert Sheldrake, "I have begun to fund what I call AI (-1) . After funding in general areas that are searching for answers I have asked people to devise an artificial method for asking questions. -my view is that intelligence has too often been modeled after computation" (EFTA00854261, OC). "AI (-1)", artificial intelligence to the power of minus one, was his coinage for a machine that would ask questions rather than answer them. In October 2016 he sent Robert Lawrence Kuhn, the host of the public-television series Closer To Truth, his "topics i intend to fund this year". The list ran: "1 the power laws. found everywhere in nature, no derivation. zipfs pareto" (power laws are the pattern in which a few items are huge and most are small, as with city sizes or word frequencies). Second came "biology and deception, mathematics and truth.? all living things are predator and prey. first line of attack and defense is deception", then power, money as "nothing more than a trusted agreement", and music as a tool to interrogate the brain (EFTA02448356, OC). Kuhn replied that several topics "fit Closer To Truth".

3.4 The 2018 "radical breakthrough" agenda

The archive's most crowd-starred topical thread (1,832 stars on Jmail) is a note Epstein forwarded to himself on 8 January 2018 under the subject "radical breakthrough". It was first seen as HOUSE_OVERSIGHT_019871. Its government duplicate is (EFTA02540965, OC; government duplicate of HOUSE_OVERSIGHT_019871). It is a numbered list of 33 questions. They include "1 beards and long hair , are meant to catch and hold smells. ?", "16. deception and its role in biollogy", "18. power lawws and probability", "24. questions = AI to the minus 1" (the PDF's text layer reads "Al", an optical-reading slip for "AI"), "29. micro bioome anti biotics" and "33 signal processing. . noise. excription". A string of names follows the list ("gromov deepak. church. martin. joscha watson, chomsky ... thiel ... bannon. barrak. ruemmler. reid"), with "membrane computing" (computation modelled on cell membranes, §1.1) and "FFT" (the fast Fourier transform, a standard way of splitting a signal into its component frequencies). A longer roll of famous names opens the forwarded copy. The note mixes real scientific questions, folk theories and name-dropping, and the same core themes recur: deception, signals, power laws, and the "AI to the minus 1".

3.5 Triad

• Ideals. The lists above, in his words: "Evolutionary Code dynamics", "Synthetic General Intelligence", "biology and deception, mathematics and truth", "AI (-1)".
• Pushback, and decisions anyway. Brockman answered the 2009 plan with a warning about the Templeton Foundation: "You will lose half the people you want by involving Templeton. The other half will go along but be unhappy" (EFTA02441292, OC). On the evening of 9 February 2013 Boris Nikolic, an investor then working with Bill Gates (his messages speak of "Bill" and of "we" working with top scientists), was advising Epstein on whom to introduce to a person the thread does not name in this part. He wrote: "I would stay away of Barnaby — I know that you are liking him but Templeton Foundation has a bad reputation of religious mambo jumbo wrapped up with pseudo science." Epstein defended Marsh: "your call, but barnaby is not like that at all. he went to oxford, only liikes real science. the princpals are wackos". Minutes later Nikolic replied that "Because of its prinicipals, that foundation has a horrible reputation. Not bad but a number of scientists are actively lobbing against", and that "It might be risky to get involved/have any connection with Tempelton at this stage". Epstein answered, "got it , who on the list do you think he would enjoy the most" (EFTA00668377, OC; EFTA02723228, OC). He gave way on this introduction but not on Templeton itself. Templeton money and Templeton-linked matching (offers to give an amount equal to a Templeton grant) run through Chapter 9.
• Supporters. Brockman's Edge circle was Epstein's main route into the scientific elite, as BuzzFeed News documented in 2019 [S32]. Kuhn took the 2016 list as programme material. Marsh replied to the 2010 list, "these are all good topics, I'm looking forward to exploring together!" (EFTA02433017, OC).

What the lists point to. The same small set of ideas recurs for a decade, so the interest was persistent rather than a passing pose. But the lists were also written for an audience: lawyers, reporters, a television host and a conference organiser. They tell us what he wanted to be seen funding, which is not the same as what he funded. Chapters 4 to 9 test the lists against the money.

4. Biology as code and deception: the centre of gravity

Argument. The idea that recurs most often in Epstein's own words is that living systems run on encrypted signals. Deception, on this view, is how organisms protect their most efficient "algorithms", so understanding life becomes a code-breaking job. Between 2012 and 2015 he turned this into repeated attempts to recruit code-breakers, including from the National Security Agency, for biology. No agency or code-breaker is shown taking it up. Energy belongs in this chapter too, because he wrote about it as free energy and entropy in living things rather than as a power-station technology.

4.1 "Truth vs deception"

The earliest full statement is a note Epstein emailed to himself on 25 November 2010 under the subject "content":

"I. biology verus mathematics, yin/yang. Truth vs deception.. the most efficient algorithm , is the most vulnerable. Therefore deception needed for protection.,, viruses, tehonology now available encryption de enctution„ wavelete anuaysism ellitcal transformations real random genetratersl. Self deception, alice — bob , well established , code theory, codebreaking, evolution butterly, computer can't deceive." (EFTA02417452, OC)

Read plainly, he is saying that mathematics deals in truth and biology in deception. A living thing that runs a very efficient routine is easy to exploit, so it must disguise itself. The tools he names are cryptographers' tools. "Alice — bob" are the stock names cryptographers give to two parties exchanging secret messages. Wavelet analysis and elliptic transformations are mathematical techniques used in signal processing and modern encryption. "Evolution butterly" points to the butterfly, the textbook case of camouflage and mimicry. The last clause, "computer can't deceive", sets up a theme that returns in Chapter 5.

4.2 Cells as encrypted senders

On 4 July 2013 he wrote to four scientists at once: Martin Nowak, the Harvard mathematical biologist whose Program for Evolutionary Dynamics he funded; Ed Boyden, an MIT neuroscientist; Neil Gershenfeld, who directs MIT's Center for Bits and Atoms; and Whitfield Diffie, a co-inventor of the public-key cryptography that secures most internet traffic. The message read:

"What I would like to explore/fund, is the premise that biological , cellular, neuronal, signals are all highly encrypted, and the possibility that each entity has its own unique encoding algorithm. Where Mathematics has no need for protection, Biology and its agents on the other hand consists of fundamentally merely helpless prey without a system that protects them from the free -energy seeking predators in their environment" (EFTA00964606, OC)

"Free energy" here is the physicist's term for the usable energy a system holds. Predators, in his picture, are hunting for it. Three days earlier he had pressed Gershenfeld on the same idea, asking for "a closer look at cellular signaling form one to another attacked the same way as other signal intelligence. the signal is chemical in nature , but that might mean electircal or physcial shape". Gershenfeld answered as a scientist would. He pointed Epstein to colleagues who study how cells sense and signal ("a colleague who's worked on pulsatile signalling", that is, signals that cells send in rhythmic pulses rather than as a steady level) and, the next day, to the systems biologist Adam Arkin ("look at Adam Arkin") (EFTA00964233, OC). In July he did not take up the framing of signals as codes. In December he answered the deception idea with a broader proposal of his own (§4.2a). The choice of addressees matters. He put an evolutionary game theorist, a brain engineer, a physicist of computation and a cryptographer on one email and asked them to treat the cell as a cipher.

4.2a A "Mathematics of Deception" retreat, 2013–2014

By late 2013 the deception idea had become a planned meeting. On 9 December Gershenfeld wrote to Ed Boyden, copying Judith Donath, Sebastian Seung, Theresa Tomic, Epstein and a "Lesley" whose surname and address are boxed: "For a Mathematics of Deception weekend retreat in sunnier climes (Jeff's island), we could all do Feb 8,9, 22,23, Mar 8,9, Apr 12,13. How are those options for you?" (EFTA01942947, OC). Epstein answered, "feb 22 23 ok if ok with everyone else" (EFTA01943390, OC). An invitation list is not attendance, and none is asserted here.

On 27 December Gershenfeld proposed a wider aim: "I haven't had a chance to return to this thread, but I was thinking that a retreat to understand deception is too modest a goal for this group. How about the more ambitious goal of building a brain? Then your interest in deception becomes a testable subsystem of the brain-building project." (EFTA01753311, OC). A subsystem here is one working part of a larger machine, a piece that can be built and tested on its own. Six minutes later by the printed times, Epstein replied: "I agree. . define -brain. - do we need a minion of them for social behavior. . I love the idea of throwing turing aside and relooking from scratch at transistors ,gates etc. .- deception and truth are asymmetrical. . like chocolate milk, : a little chocolate makes it chocolate milk, a few lies makes one a deceiver, where a few truths do not make one a truthteller." (EFTA01944659, OC). "Throwing turing aside" refers to Alan Turing's model of computation, the step-by-step machine that underlies ordinary computers. Transistors and gates are the tiny switches from which computer chips are built. On 31 December Gershenfeld told the group: "Jeff and I discussed a potential larger focus for the weekend: planning a build-a-brain project. That of course has many potential meanings, but I like the challenge, and within that deception would be a subsystem to design or emerge. My personal interest would be the physical computational substrate for consciousness." (EFTA01751277, OC). A computational substrate is the physical material that a computation runs on, such as silicon chips or, in a brain, neurons.

The meeting moved to New York. By 11 February 2014 the group had settled on 14 March, and Gershenfeld asked [R-22], "Do you want to book travel, or should we do that and you reimburse?" (EFTA02114973, OC). On the day itself Boyden withdrew: "My apologies -- for the Obama BRAIN initiative, I need to work at MIT today" (EFTA01935419, OC). Lesley Groff then wrote that "Jeffrey prefers to either reschedule for April in NY or he could come to Boston possibly sometime April 23-25...Jeffrey was flying in for the dinner so he prefers to reschedule". [R-24] proposed "continuing with this evening's plan - and have Ed videoconf in", and Seung agreed: "four out of five isn't bad" (EFTA00372888, OC). The documents checked for this paper do not show whether the dinner took place, whether it was rescheduled, or whether the island retreat ever happened.

4.3 "A guy from nsa": the code-breaking plan, 2012–2015

The plan to bring in professional code-breakers ran for at least four years. The checks for this paper found it in eleven separate government originals.

January 2012. Writing to Tom Pritzker, the Hyatt hotels chairman, who was on his way to Washington, Epstein asked: "ask her if we can nsa code breaking ( signal processing) to work on brain actintiy and intracelluar deception". Pritzker had just written "Shit. I'm seeing Darpa lady tonite". DARPA is the Pentagon's research agency, and the woman is not named in the email. Pritzker then asked, "Huh? Explain a bit more" (EFTA01845817, OC). Epstein explained: "think of communication between cells in the body made up of proteins„ they are merey signals that we cannot yet decode NSA decodes signals .. in many languages and math forms.. the algoritjms should be tested against biologically generated signals. it is very clever and I promise a leap in brain understanding" (EFTA00928089, OC). No further reply from Pritzker on the subject was found.

March 2012. He asked Boris Nikolic to set up a conversation about "signal processing , and its application to biological systems. wavelet anaylsis, elliptical transformations, either her or if you can help arrange through her the NSA guys" (EFTA00930323, OC). The woman is identified in the email only by a first name, which is not expanded here.

August 2012. To the banker Maurice Sonnenberg he wrote that America's intelligence agencies "could be put to work on breaking the dna code or protein signal problems. breaking foreign codes is the expertise of the us and nsa. it would be great to know which agency button to push" (EFTA01871811, OC). Three days later he sent the instruction that became the template for his public image: "I want to use signal intelligene , and cryptanalysis to understand the signals in biologcal systems„ cell to cell. dna to growth„ , virus protection and operation. put it out on the blog" (EFTA01872375, OC). Signal intelligence is the spy agencies' craft of intercepting other people's communications. Cryptanalysis is the art of breaking the codes those messages are written in. Ten days later the writer who ran his websites reported having "drafted and fleshed out seven forum concepts including: signal intelligence & biological systems, music and the brain, dreaming, alternative currencies and the predator/prey dynamic of biology" for the site jeffreyepsteinforum.com (EFTA01890262, OC). The blog instruction was carried out. The scientific recruitment, as far as the record shows, was not.

December 2012. In a questionnaire for a blog post he listed his "work this year": "1. do dreams occur 24 hrs, and conciousness come on and off 2. signal intelligence for biological communication ... 4.biology, vs mathematics, decetion vs truth" (EFTA02564253, OC).

August 2014. Writing to Bill Gates, he described a new venture: "I am starting an institute at MIT calling it ONE SCIENCE . basically, to explore the intersections of what appears to be diverse fields. Signal intelligence, toplology, big data, mathematical biology, neuro etc. It will be run by Joi Ito. head of media lab" (EFTA00995604, OC). Topology is the branch of mathematics that studies shape and connection. No institute of that name appears in MIT's public report on Epstein [S04]. The email is evidence of what he told Gates, not of an institute.

April 2015. A long note to himself included: "why not attack the bio language the same way as speech. have nsa work on decrypting the differnce in signals from a cancer cell as opposed to a helathy one. we know that cells send out messages , am i in the right place" (EFTA02501778, OC).

September 2015. He asked Kathy Ruemmler, a former White House counsel then in private practice, "can you find a guy from nsa that can think about signal intelligence applied to DNA. I woiuld love to get some contact with decryptors that can work on biology „ I will fund. I want to intercept communicaiton between living cells. in organisms , the rules should be the same for language". Ruemmler replied: "I will make some inquiries" (EFTA02357356, OC).

October 2015. To Joi Ito, director of the MIT Media Lab, he floated a "collider" idea for President Obama and called it "not as good as bio signal intelligence" (EFTA02340311, OC).

4.4 Energy, entropy and "default on"

The same April 2015 note opens with a theory of living things as energy systems: "unlike machines that turn on when the switch is thrown. . in essece default off. biological systems are default on. they attempt to stay on." It moves on to consciousness as a "screen" in front of dreams that are "ALWAYS on", a screen that "takes energy" to keep opaque (EFTA02501778, OC). Entropy, the physicist's measure of disorder, which every living thing must keep pushing back, is the background to all of this. In 2010 he told Marsh that "there are no energy equations in predator prey models", the equations ecologists use for how populations of hunters and hunted rise and fall (EFTA02433017, OC). Energy in the technological sense hardly appears. There is no fusion-power funding thread: the scan found only four distinct statements on fusion, two of them his. The one striking line is a boast to Al Seckel in October 2009: "regarding cold fusion. i killed pons years ago" (EFTA00884555, OC). Stanley Pons was one of the two chemists whose 1989 claim of room-temperature fusion collapsed when other laboratories could not repeat it. Epstein's claim to have "killed" the idea is his own and has not been checked against any outside record.

4.5 Triad

• Ideals. In his words: cells send "highly encrypted" signals; "the rules should be the same for language"; "I will fund"; "put it out on the blog".
• Pushback, and decisions anyway. Nobody in the record rebuts it outright. The nearest thing is Gershenfeld's reframing of deception as "too modest a goal" and a "testable subsystem" of a brain-building project (§4.2a). Goertzel wrote in 2026 that Epstein "once proposed that “deception” was the core of all intelligence and of molecular biology" and that "I thought he was straining things and said so" [S65]. That is his later account; no email of the time was checked for it. Otherwise the responses are puzzlement (Pritzker: "Huh? Explain a bit more"), deflection (Ito: Obama did not seem "that interested in hard sciences") or polite compliance (Ruemmler: "I will make some inquiries"). The record does not show any intelligence-agency person, DARPA official or code-breaker engaging. This absence was checked: every email in the archive was searched for "nsa", "decryptor", "cryptanalysis" and "signal intelligence" within 300 characters of "dna", "cell", "biology" or "genome", and the 82 distinct hits were read (gaps_phase2.md). The decision he made anyway was to put the idea into his public profile through the blog, and later through a draft character reference written about himself (Chapter 10).
• Supporters. The scientists on the 2013 email did not, in the record, take up the encryption premise as stated. In July Gershenfeld redirected it toward ordinary cell-signalling research. In December he went further and organised a "Mathematics of Deception" retreat, with Boyden, Seung and Donath among those invited (attendance is not asserted), and proposed making Epstein's "interest in deception" a "testable subsystem of the brain-building project" (EFTA01942947, OC; EFTA01753311, OC). The work he funded did overlap with it in places. Ben Goertzel offered to build a "deception character" with his OpenCog software (Chapter 5), and Martin Nowak's evolutionary dynamics, the mathematics of how strategies spread in populations, is the scholarly field closest to his predator-and-prey language [S51, S52].

Where his idea meets mainstream biology. Biologists do study cell signalling, the chemical messages cells use to coordinate growth, defence and death. Evolutionary biologists also study deception, in mimicry and camouflage, and in the "honest signals" that animals cannot easily fake (a phrase that appears in his 2018 agenda). What mainstream biology does not do is treat cellular signalling in general as encrypted, as messages hidden from a listener by design. That step was Epstein's own. It explains why he kept reaching for spies rather than biochemists, and why the people he approached answered with puzzlement, referrals or a broader counter-proposal, rather than code-breaking research.

5. Machine minds: scepticism, robots and cheap AGI patronage

Argument. Epstein mocked the idea that the brain is a computer. Yet for years he paid small sums to people trying to build artificial general intelligence. He imagined machines that would deceive, and a robot built to his own specifications. The money was small and early, and what it bought was mostly access and a story to tell about himself.

5.1 The sceptic: "as silly as artificial intelligence"

In June 2015 Epstein and Noam Chomsky, the MIT linguist, argued by email over a paper Chomsky had sent him, and the exchange shows both his scepticism and how it was answered (EFTA00854924, OC). Epstein opened by saying that the paper's "mathematical descriptions lack rigor" and that it suffered from describing things "using the engineering metaphors or the tools of the moment.ex. The human or its brain as a computer". Chomsky's "Merge", the single operation of combining two items into a set that sits at the core of his theory of grammar, was in Epstein's words "quite naive and unfortunately incorrect". Chomsky answered that he loved criticism, "But in this case, I just don't see the arguments". The papers, he said, assume that language in the mind "is a computational system ... easily formalized. I don't know of any coherent alternative." Epstein came back with "brain as a computer? as silly as artificial intelligence.," and a picture of choice: "The simple example I teach re quantum is when i try to decide should i order fish or meat ... both choices like schroedingers cat exist as a superpostioned wave function that collapses and a choice is made. . yes i am taking liberties". Schrödinger's cat is the famous thought experiment in which a quantum object is in two states at once until it is observed. Chomsky agreed that for choosing an action "the computer model is not helpful at all", a view he had long held, but added, "I don't frankly see how the Schrodinger analogy helps in this case." He also corrected Epstein's description of people who attended his seminars: they "weren't students working on computers. Rather, professional mathematicians and physicists." When Epstein said the computer model of living systems "has not led to many conherent theories", Chomsky replied that it "is often misused, but it's often used quite effectively", for language and "to study insect navigation and much else". The same month he told Rupert Sheldrake that "intelligence has too often been modeled after computation" (EFTA00854261, OC). In May 2016, writing to Larry Summers, he dismissed the big machine-intelligence efforts of the day, "deep mind. blue brain. paul aliens brain institute , watson, ( ibm ), all. engineering attmpts. . Its the e incarnation of cybernetics. from the 50s" (EFTA02463802, OC). Cybernetics was the 1950s science of feedback and control in machines and animals. In 2018 his "radical breakthrough" list still had "questions = AI to the minus 1" (EFTA02540965, OC; government duplicate of HOUSE_OVERSIGHT_019871; see §3.4).

5.2 "A computer that will lie to me"

Because he believed deception was the signature of life (Chapter 4), he drew the obvious conclusion for machines. On 29 August 2013, after a sushi dinner with an unnamed correspondent (the To line of the original is blank), he wrote: "for the agent to protect itself it needs Alice and Bob like behavior and all its permuations. . protetcion from viruses , it might engage the same methods . add junk, encrypt, have third party verification. etc. Unfortunately computers don't lie they merely generate some errors. I want build a computer that will lie to me. ( a female turing machine )" (EFTA01970128, OC). A Turing machine is the mathematician Alan Turing's abstract model of a computer. "A female turing machine" is Epstein's own gendered joke: in his framing, lying is what would make a machine female. The line is short, but it joins his two recurring preoccupations, deception and women, in one sentence.

5.3 "An intellignet robot ... what she should look like"

On 20 January 2013 he wrote to a recipient whose name is redacted in the original ([R-6]): "I am builidng an intellignet robot. really, i am trying to decide what she should look like? age body type. its a high level project „ you can you tube some of it on david hanson robotics" (EFTA01906627, OC; see J1-4). Hanson Robotics is the company later known for the humanoid robot Sophia. Jmail shows a first name as the recipient, but the government original has a redaction box in the To line (Table J1, row J1-4). In 2019 a journalist reported that Epstein had told him he funded Sophia [S43]. The checks for this paper found no payment to Hanson Robotics and no Hanson reply in the archive. The email shows what Epstein said he was doing, and the specification he cared about: a female robot of a chosen "age" and "body type".

5.4 Paying for AGI: Goertzel, OpenCog and the Singularity Institute

The actual money for machine intelligence went mainly to Ben Goertzel, an American AI researcher who leads the open-source OpenCog project, an attempt to build artificial general intelligence. The flows were small and spread over years:

• April 2009. Goertzel asked whether Epstein would pay $2,000 to record AGI Summer School lectures. A check "should be sent to the Singularity Institute for AI (a nonprofit)". Epstein's forward to Richard Kahn, his accountant, reads "Send 5k" (EFTA00663702, OC). The Singularity Institute, now the Machine Intelligence Research Institute, was then a small charity built around the idea of a coming "singularity", a point at which machine intelligence outruns human control.
• September 2009. "is there a 501 c3 that i could give the 50k to/?" Goertzel: "Yes: the Singularity Institute for AI [address redacted] ... which is where your donation to AGI-09 was routed". Epstein: "please send details i will send 50k monday". Goertzel: "The AI we build will thank you one day!" (EFTA00770109, OC). The message's From line reads "[R-17] on behalf of Ben Goertzel"; it is signed "ben". A 501(c)(3) is a U.S. tax-exempt charity. AGI-09 was the second conference on artificial general intelligence.
• June 2011. Goertzel asked again for "financial support of OpenCog HK project" and mentioned that he had covered a $10,000 payment himself. Epstein wrote "10k" on the forward to Kahn (EFTA00911521, OC).
• December 2013. Goertzel asked for $45,000 as the "10% corporate donation" that a Hong Kong government grant required, routed through the charity Humanity+, and added, "I want to see the deception character too, and I look forward to helping my team create it using OpenCog ;)". Epstein to Kahn: "send 45k" (EFTA00978444, OC). Later that month Goertzel asked Kahn when "Jeffrey's generous grant of $60K" would reach his company, Novamente LLC (EFTA00634051, OC).
• January 2015. On 24 December 2014 Kahn asked Goertzel "whom we shall direct funds to for attached", and on 4 January Goertzel asked: "has Jeffrey decided to fund this proposal?" On 6 January Kahn replied: "Due to the current environment with all the bad press Jeffrey is receiving we are going to have to put this funding on hold". Goertzel then asked "if Jeffrey might still be able to help with $25K for my "corporate contribution" to the OpenCog Hong Kong project. He has helped in this way every year since 2010, usually via a donation to Humanity+." Epstein's reply reads, in full, "yes 25". The top message of the chain, signed "ben" (its From line is boxed, [R-21]), thanks him and proposes to "resume conversation btw you, me and Joscha on the larger proposal" (EFTA02373775, OC; see J1-12). Goertzel forwarded the reply to Kahn ("OK that is great news if he can help with the $25K now"). The next day Kahn wrote to Epstein, "confirming ok to send ben 25,000", and asked "if we should send from Enhanced or STC". Epstein answered "enhanced" (EFTA00867122, OC; EFTA00867128, OC). Business Insider identifies STC as Epstein's Southern Trust Company [S64]. The e-mails do not spell out "Enhanced". "yes 25" names no unit or purpose, and the $25K and 25,000 readings are Goertzel's and Kahn's. No payment record was seen. The printed times in this chain mix time zones (Epstein's reply is printed at 6:54 PM, earlier than the 7:26 PM message it answers), so the order is taken from the quoting (J1-11). Goertzel's reply to Kahn about the allegations is discussed in §11.4.

An undated proposal, its title page naming "Ben Goertzel and Gino Yu, Principal Investigators" (Yu is an academic in Hong Kong) and marked "Rough draft, not yet for distribution", is headed "OpenCog AGI Toddler Project" and 'Open-Source "Human Toddler Level" AGI by 2015 for US $3M Total' (EFTA01114145, OC). The PDF shows no sender, recipient or date. No payment of that size was found. In 2014 Goertzel told a writer preparing an article that, "technically", the foundation had not funded his Ethiopian lab, and said he did not mind a press release associating them. That episode belongs to Chapter 10.

Joscha Bach, a German cognitive scientist who studies how minds could be built, was the other main beneficiary. MIT's report found that $300,000 of Epstein's Media Lab gifts in 2013 and 2014 paid for Bach's research, and that the Lab hired him "in large part because Epstein subsidized the cost" [S04].

5.5 Arguing with Chomsky about meaning

Epstein used his access to argue with the scientists he funded or courted. In July 2016 Bach presented an argument about learning at a meeting with Chomsky. Afterwards Bach wrote to Epstein, copying "Martin" and Moshe Hoffman: "thank you for your support and encouragement, even where I fail. Sorry for being such an embarrassment today", adding that Chomsky was "very generous, patient, kind and humble on the personal level, even though he did not feel like conceding anything on the conceptual one". The same message went on to a claim about race: "In the US, black children outperform white children in motor development, even in very poor and socially disadvantaged households, but they lag behind (and never catch up) in cognitive development even after controlling for family income" (EFTA02458509, OC; a message contained in the House Oversight chain HOUSE_OVERSIGHT_025954; letters lost to encoding on the page, as in "=ouseholds", are restored). Epstein's reply to this message is the "ubermensch" email discussed in §8.1. On 26 December 2016 Epstein put his own idea to Chomsky: "has anyone in the past made a linguistic model that used fields instead of definitions. similar to a magnet field. to some degree. there is no one vector that describes the word but a field of them ... not sure it is useful. but having fun". Chomsky answered the same day. The closest thing he knew was "an array of meaning postulates, in Carnap's sense ... Multidimensional, but not a field. I don't see how to work it out, or to extend it to the interpretations of longer units". Later he added, "Still don't see how to work out field models" (EFTA02664783, OC; government duplicate of HOUSE_OVERSIGHT_030328). Meaning postulates are the philosopher Rudolf Carnap's rules linking words to other words. Today's language models do represent words as points in spaces with many dimensions, built from the contexts in which the words appear. Epstein's "field" is a loose cousin of that idea, but nothing in the record shows that he developed it beyond this email.

5.6 Triad

• Ideals. "AI (-1)", a machine that asks questions; "a computer that will lie to me"; "an intellignet robot" (his spelling); and his scorn for "engineering attmpts".
• Pushback, and decisions anyway. Chomsky's replies are the clearest intellectual pushback in this strand: courteous, specific and unconvinced. He defended the computational view of language that Epstein called "silly" ("I don't know of any coherent alternative"), rejected the quantum analogy, and later could not see "how to work out field models". Bach reported that Chomsky conceded nothing. Epstein's response was to keep funding his own picks. He promised $50,000 for AGI-09 ("i will send 50k monday"), told his accountant to "send 45k" for OpenCog Hong Kong, and, through the Media Lab, paid for Bach's research ($300,000, by MIT's count [S04]). In January 2015 his own office put an OpenCog payment "on hold" because of "all the bad press" (EFTA02373775, OC; see J1-12). He then replied "yes 25" to Goertzel's renewed request, and when Kahn asked "confirming ok to send ben 25,000" and whether to use "Enhanced or STC", he answered "enhanced" (EFTA00867122, OC; EFTA00867128, OC).
• Supporters, then and later. At the time Goertzel was effusive ("The AI we build will thank you one day!"). In February 2026 he published "Goertzel vs Epstein", his account of the relationship [S65]. Press reports that month described the funding and an introduction that helped him in Hong Kong [S63, S64]. Bach thanked Epstein for "support and encouragement" in 2016. In November 2025 he published "On the Jeffrey Epstein Affair", his statement on the relationship [S19], after the Boston Globe reported on his emails [S18]. WIRED in 2025 described the 2002 St. Thomas meeting on human-level AI that Epstein hosted with Marvin Minsky [S40, S41].

What the money points to. Machine AI was a small, early and cheap part of the portfolio: thousands to tens of thousands of dollars at a time, plus about $300,000 for Bach through MIT. Even these gifts were framed through deception (the "deception character") and through his scepticism of computation. He funded people more than programmes. Some of them, Goertzel above all, told him what he wanted to hear. Others, like Bach and above all Chomsky, argued with him.

6. Quantum minds and the Penrose–Hameroff circle

Argument. From 2014, and especially in 2017, Epstein's attention moved towards quantum theories of consciousness and quantum computing. He committed money to the consciousness conference before knowing the details. The leading quantum-consciousness theory is disputed by most physicists and neuroscientists. The record shows no scientist warning him off it, which is itself informative.

6.1 "In for at least 50k, before knowing anyhing"

Stuart Hameroff is an anaesthesiologist at the University of Arizona who runs the Center for Consciousness Studies and its long-running conference, "The Science of Consciousness". With the Oxford mathematician Roger Penrose he proposed "Orch-OR" (orchestrated objective reduction). The theory holds that consciousness arises from quantum effects inside microtubules, the tiny protein tubes that act as scaffolding in brain cells [S36]. On 30 March 2017 Hameroff told Epstein that the 2017 conference in Shanghai had collapsed after a dispute with its Chinese partner, who had promised his centre "$100,000 per year for 5 years". The Center had already "put out 35 to 50k for speakers (Including 9k for Roger Penrose business class from London)". Epstein replied: "who are the speakers and what is the cost. Im in for at least 50k, before knowing anyhing" (EFTA01050783, OC). In the same chain Hameroff explained the move to San Diego, "fortuitously where the Penrose Institute will be formed" (EFTA00697075, OC). Two days later Epstein asked Deepak Chopra, the alternative-medicine author: "Should I send some small support to penrose? I see you are scheduled there" (EFTA02654432, OC). In May 2017 Hameroff asked whether to "reserve a room at the Hyatt for you", and Epstein forwarded the Penrose Institute thread to Masha Drokova, a venture investor in his circle (EFTA01045402, OC).

6.2 The Penrose Institute

James Tagg, who was helping to organise the Penrose Institute, kept Epstein informed. On 20 February 2017 he wrote, "We are making progress on the Penrose Institute and have now found a person to lead the physics side ... the first demonstration of brain tissue preserving entangled photons was demonstrated by a team in New York, not a proof of any quantum effects in the brain but an important enabler". Entangled photons are pairs of light particles whose properties stay linked however far apart they are. Epstein's question was practical: "how is the funding coming? sources amounts? etc" (EFTA02366785, OC). A 2016–17 prospectus for the Institute listed "Human and Artificial Intelligence", a "Quantum Biology Lab", "New Physics" and "Aging and Cognitive Longevity" (EFTA00606986, TP). No payment by Epstein to the Penrose Institute was found in the emails. Jmail also shows staff emails about a personal gift for Penrose in June 2017. These were not checked against the originals, so they are treated only as a lead.

6.3 Quantum computing: the Lukin day

Quantum computers compute with quantum states rather than ordinary on/off bits, and for some problems they could be vastly faster. On 23 October 2017 Drokova told Epstein: "just spoke to Misha Lukin. you're meeting him at 11:30am tomorrow -- Martin Novak connected you ... I think quantum computer is not existing yet and I think Misha has the highest chances to build it. he is brilliant" (EFTA02363121, OC). Mikhail Lukin is a Harvard physicist and a leader in building quantum computers from arrays of single atoms. A staff-prepared schedule lists "Appt w/Misha Lukin" among meetings at Martin Nowak's office (EFTA00285408, CONFLICT; see J1-2). Jmail heads the block "Mon. Oct. 23, 2017 NY (cont.)"; the original has a redaction box there ([R-3]), and by page order the block falls between the 23 and 25 October pages (Table J1, row J1-2). The page also carries redactions [R-4]. In January 2018 Drokova offered to introduce "my ex boss Serguei", who "founded Russian and London quantum center", describing him as Lukin's "classmate/friend/sponsor" (EFTA02540352, OC). Two independent documents, Drokova's email and the staff schedule, agree on the meeting and on Nowak's part in it. Nothing in the record shows Epstein funding Lukin's work. As far back as December 2009 he had described himself as "funding a project on quantum computing at mit" (EFTA01822876, OC). MIT's report found $225,000 of post-conviction gifts to the physicist Seth Lloyd, who works on quantum computing [S04].

6.4 What the physicists say

Orch-OR has strong critics. In 2000 the physicist Max Tegmark calculated that quantum states in the brain would break down, or "decohere", in between a ten-trillionth and a hundred-quintillionth of a second, far faster than nerve cells work. He concluded that the brain "should be thought of as a classical rather than quantum system" [S91]. Hameroff and colleagues replied in 2002 that Tegmark had modelled the wrong process and that protected states could last longer [S92]. In 2006 the neuroscientist Christof Koch and the physicist Klaus Hepp reviewed the field in Nature and found no evidence that brains rely on quantum coherence [S93]. Penrose shared the 2020 Nobel Prize in Physics for work on black holes, not for Orch-OR [S96]. The theory remains a minority position, though one with prominent defenders [S35, S36].

6.5 Triad

• Ideals. "Im in for at least 50k, before knowing anyhing"; "Should I send some small support to penrose?"; "funding a project on quantum computing at mit".
• Pushback, and decisions anyway. No one in the archive warned Epstein off Orch-OR or the Penrose Institute. The mainstream critique existed in the scientific literature [S91, S93], but no email found puts it to him. (Coverage: all 1,755,614 non-promotional emails were searched for "decoher", "Tegmark", and words such as "wrong", "nonsense", "pseudo" or "sceptic" within 300 characters of "Orch-OR", "Penrose", "Hameroff" or "microtubule". The 82 matching documents are guest lists, programme links and a Chopra–Tegmark debate video, not critiques.) He committed money before seeing a programme.
• Supporters, then and later. Hameroff, Tagg and Drokova brought him in. Hameroff was later reported by the University of Arizona's student newspaper as one of two professors named in the files. The paper quotes his post of 3 February 2026: "I deeply regret interacting with Jeffrey Epstein in 2017 and strongly condemn his awful activities. He offered to help financially with a conference problem and I accepted. Nothing more" [S66]. No public statement by Roger Penrose or by Mikhail Lukin about Epstein was found (web searches on 11 October 2026 covering general news and the main file-aggregator sites; gaps_phase2.md).

What it points to. The quantum strand fits a pattern visible elsewhere. Epstein was drawn to theories that promised one grand key to mind and matter, and he funded the people around them quickly, informally and in modest amounts. The physicists' disagreements show up in the published literature, not in his correspondence.

7. The genome chapter: George Church, his own DNA and longevity

Argument. Biodata, meaning information read from a body (genes, cells, microbes), was where Epstein's talk turned into transactions: sequencing his own genome, a company to invest in George Church's list of genetics ventures, and longevity medicine. Church's caution at the time and his public apology later make this the clearest case of a participant's framing changing between then and now.

7.1 Sequencing Epstein

Genome sequencing means reading out the full sequence of a person's DNA letters, about three billion of them. On 24 February 2014 Joseph Thakuria, a physician-researcher at Massachusetts General Hospital, priced it for Epstein: whole-genome costs "through Illumina which are $9500 for a normal turn around time and $11,400 for 'rapid turn around time'", adding that "The rapid turn around time isn't necessary in Jeffrey's case". Illumina is the dominant maker of sequencing machines. Epstein's assistant Lesley Groff relayed "Jeffrey's reply": "lets start and do the full genome with rapid turn around. 11, 400". Thakuria later noted it was "pending approval through Partners/MGH IRB for my formal research study" (EFTA00650143, OC). An IRB (institutional review board) is the ethics committee that must approve research on human subjects. In this case Epstein's own genome was to go through a research study. When Thakuria's note was forwarded to him, Epstein told Groff and his accountant Richard Kahn, "give him 10k". The chain, which runs to May 2015, ends with his bookkeeper attaching a $2,000 check to Thakuria issued in June 2014, and Groff adding, "I think we sent along 10k as well" (same document).

7.2 Church's "$10M" list

George Church is a Harvard geneticist and one of the pioneers of modern genome sequencing and gene editing. In November 2011 Epstein asked him, "did the cloning isssue , give you pause?" Church replied: "Yes. I'm working toward this goal fairly rapidly, but trying to do so in a way that minimizes risk to the field. James Wilson (who happened to be my physician, in the 1990s) set back the field of gene therapy by a decade by rushing" (EFTA01850962, OC). Gene therapy means treating disease by adding or changing genes in a patient's cells. The email does not say which "goal" Church means beyond the cloning question it answers, and this paper does not read more into it.

In August 2014 Church sent "a first pass list of potential investments for the $10M". It included "$200,000 Supercentenarianstudy.com" (genomes of people who live past 110), "$1,500,000 eGenesisBio.com: Using CRISPR to engineer animals for resistance to infectious diseases and to act as transplantation donors for major organs", "$800,000 Aging reversal" and "$200,000 Cold-resistant elephants". CRISPR is a molecular tool for cutting and editing DNA at a chosen spot. Epstein replied, "lets start on first investment". He proposed forming the investment company with his lawyer, Darren Indyke, copied: "I like george gaage, if ok with you I will form and get documents to you". On 25 August he wrote, "co formed not we have to invest" (EFTA00995805, OC). In September Church explained the supercentenarian business plan: "to patent products based on genomic (and epi-genomic) variants which are unique to one or more supercentenarians ... The CRISPR is used for changing DNA in human cells ... to match the SC variant" (EFTA02387794, OC). Epigenomic means the chemical marks that switch genes on and off without changing the DNA letters.

The emails show a list, an agreement to start and a company said to be formed. They do not show what, if anything, the company invested. No transfer record for it was found in the archive's bank documents.

7.3 Longevity: "protecting yourself FROM nature"

Longevity medicine tries to slow ageing itself. In June 2015 Epstein told Peter Attia, a physician who writes on longevity, "conjecture - natural . is good when you are young. natural is bad when you are old. . proecting yourself FROM nature is the essence of longevity. your metformin is probably right on" (EFTA02353720, OC). Metformin is a cheap diabetes drug that some researchers are testing as an anti-ageing treatment. Attia replied that his goal was "to combine the best evidence of diet/sleep/exercise/drugs/hormones into a super longevity package".

7.4 Triad

• Ideals. "lets start and do the full genome"; "lets start on first investment"; "proecting yourself FROM nature is the essence of longevity".
• Pushback, and decisions anyway. Church's 2011 reply is caution, not refusal: move "in a way that minimizes risk to the field". Epstein pressed ahead with the 2014 company.
• Supporters, then and later. Church supplied the list, Thakuria ran the sequencing, and Groff and Kahn handled logistics. In 2019 Church apologised for his contacts with Epstein, attributing them to "nerd tunnel vision" [S49]. BuzzFeed News reported the wider pattern of scientists who took Epstein's money after his conviction and their later statements [S50]. No public statement by Thakuria was found.

What it points to. The genome strand is where Epstein's interest in "the code" of life met real laboratory work. It is also where the interest turned personal: his own DNA, his own ageing. Chapter 8 follows the same line to its far edge.

8. Designer babies, melding and immortality: the far edge

Argument. At the far edge, the archive shows Epstein engaging with schemes to remake humans. The sharpest case is Bryan Bishop's 2018 "designer baby and human cloning company". Epstein wrote that "the problem is only if i am seen to lead". The word "deniability" first appears in a message from Austin Hill ("total deniability") and then in Bishop's ("sufficient level of deniability"), and Bishop wrote of "absolute anonymity", giving his own reason for it. Epstein also asked for a legal reading first. This sits alongside his "ubermensch" talk about merging the best of human types, and his scepticism about mind uploading and cryonics. No payment to Bishop has been found, and Bishop says he never took Epstein's money. The chapter is about intent and engagement, not a funded project.

8.1 "Ubermensch could be the melding of humans"

In July 2016, writing to Joscha Bach after the meeting described in §5.5, Epstein said: "what i like is the idea that ubermensch could be the melding of humans, put together in one brain an increased motor system, an increased awareness , an increased status calculator. , ( blacks jews, women ) ubermensch could be the combination of the best of humans not the best of a specic race or gender. fun idea. we can call it super intelligence. or uberintelligence" (EFTA00824159, OC). The order matters. Epstein was answering Bach's message sent in the early hours of 10 July (§5.5), which had said that in the US "black children outperform white children in motor development ... but they lag behind (and never catch up) in cognitive development even after controlling for family income" (EFTA02458509, OC; a message contained in the House Oversight chain HOUSE_OVERSIGHT_025954). The race claim entered the thread in Bach's message; Epstein's bracket picked it up. The Übermensch, or "overman", is the philosopher Friedrich Nietzsche's figure of a human being who surpasses ordinary humanity [S21]. Epstein's version is a fusion of supposed group traits into one mind. He presents it as anti-racist ("not the best of a specic race"), but it rests on the racial and sexual stereotypes in the bracket.

Two weeks later the thread returned to the same claim. In a message timed 3:02 AM on 23 July 2016, Bach quoted an earlier line of Epstein's: "you might be able to make blacks smarter by changing the time for motor layer development and changing the time for other layers". Bach answered, "Exactly. I looked up the statistics, black kids in the US have slower cognitive development (and never catch up), which the study of course attributed to social factors without any evidence", and went on to speculate about Europeans' supposed adaptations to agriculture and about gender differences, which he thought "mostly motivational" (EFTA00823256, OC; a message contained in the House Oversight chain HOUSE_OVERSIGHT_025954). The government copy prints Epstein's quoted lines and Bach's answers interleaved inside Bach's message, without quote markers; they are attributed here by their position and style. Jmail splits a later message in the same chain differently from the government original (Table J1, row J1-6). Mainstream genetics rejects attempts to rank human groups by innate intelligence [S45], and the developmental difference Bach cited has long been tied to environment and care practices [S84, S85]. Bach addressed these emails in his 2025 statement [S19].

In April 2019 Epstein sent Martin Nowak a note on "the axioms of miracles ... the dark matter of the mind. we know its there but cant see it measure it, mathematzie it. but we feel it. intuit it ... its women science . larry summers was right women are not good at mahtmatics they are good at miracles. giving birth for example" (EFTA02634697, OC). The reference is to Summers's 2005 remarks about women in science.

8.2 Upload, connectomes and cryonics: the audience, not the author

Mind uploading is the idea of copying a mind into a computer. A connectome is a complete wiring map of every connection in a brain. Cryonics is the freezing of the dead in the hope of future revival. Epstein was more sceptic than enthusiast. In May 2015 he told Dmitry Itskov, the Russian entrepreneur behind the "2045 Initiative" for uploading minds: "since 1990. I have funded attempts at MIND BRAIN BEHAVIOR, providing the intial moneys for thefirst Harvard commitee on the subject. Initial FFunding of the santa fe institue. . 25 years later. no coherent theory ... frankly not very much to show for it" (EFTA00856152, OC). When he asked Bach in 2018 about Nectome, a start-up proposing to preserve brains for future uploading, Bach answered that "chances that the current plan for restoring memory from connectome works one day are quite close to nothing but not absolute zero" (EFTA02508072, OC). In 2015 Bach had weighed cryonics for him as a bet: "$500 a month for a 1% chance of immortality" (EFTA00847724, OC).

8.3 The Bishop proposal, July–November 2018

Bryan Bishop is a bitcoin software developer whose do-it-yourself "designer baby" project MIT Technology Review described in 2019 [S39]. A designer baby here means a child whose genes are deliberately altered at the embryo stage. Because the change is made so early, it is copied into every cell of the child and passed on to the child's own children. Scientists call this germline editing, and many countries prohibit using it to start a pregnancy.

Times in this thread are printed in the originals without time zones, and the stored copies mix zones, so the order below follows how the messages quote one another, not the clock times.

The approach (20–21 July). The thread began with Austin Hill, a co-founder and former chief executive of the bitcoin company Blockstream. On 20 July, after a call, Hill wrote: "Fun talking today. Lots of good ideas and ways to follow up. On idea stuck in Bryan's head - and if your serious ... Then we should setup some secure communications and I'll let Bryan pitch you on a way to have total deniability." Epstein answered "ok". The next day Hill asked: "How good is your info sec / serveillance teams - do we need to bring some skills to the table to do a proper info sec sweep? We've got in our network guys whom did Snowden's escape to Russia level counterseivllance. Also - should we setup or assume you've got OpenPgp capabilities for secure email?" Information security ("info sec") means protecting communications and data, and OpenPGP is a standard for encrypting e-mail. Epstein answered "Later". Hill then wrote: "I working with ideas for Bryan. I know you've hid you're investments through Joi in the past, but the way this might work might be an offshore hedge fund, investment trust. That's makes a few acquisitions of Bryan's and mine side projects. All settled and managed in Tumbled bitcoin transactions." An offshore fund is one registered outside the investor's home country; a hedge fund is a privately run investment pool; an investment trust is a company that holds investments for its shareholders. "Tumbled" bitcoin is cryptocurrency passed through a mixing service to obscure its origin. Epstein answered: "I dont hide" (EFTA01004727, OC). Hill's remark about investing "through Joi" is taken up in §9.4.

The pitch (21–22 July). Later on 21 July Bishop sent a deck "about my designer baby project". He asked about Epstein's "requirements for secrecy and privacy, specifically regarding reputational risk and also any financial involvement", and suggested doing the work "under the banner of my designer babies project, since the other deliverable is similar and shares so many of the same procedures and lab requirements. This might offer a sufficient level of deniability". The "other deliverable" is not defined in the email, and this paper does not interpret it. Epstein replied: "im traveling in mid east until the 1st. lets do it after that , i have no issue with investing the problem is only if i am seen to lead". Bishop proposed a call ("How about Wednesday August 2nd @ 4pm EST?") and continued: "Sounds like you could be the first investor as long as absolute anonymity is kept. I have always envisioned there would be anonymity requirements about babies-- we can't publicly identify who these are or their parents or benefactors, it would brand the child as (essentially, and sadly) a freak for life in the media. So I had always assumed the investor would need absolute anonymity, for these kinds of products." He added that "Austin has been advising me on this" (EFTA01004839, OC; see J1-9). Epstein added, "i cant program but iam not bad at structuring :)" (EFTA01004841, OC).

The call and the budget (2–5 August). Bishop's Google Calendar invitation, sent on 23 July with Bishop as organiser, set a call for Thursday 2 August 2018 at 4 pm (his first proposal had named "Wednesday August 2nd", but 2 August 2018 was a Thursday). A calendar alert in Epstein's account reads: "Designer babies (Bryan Bishop) Agenda- Designer babies, structuring, finance, privacy, project goals, feasible outcomes, timelines" (EFTA02598225, OC). Epstein also forwarded the invitation to Groff (EFTA00478987, OC). The alert shows the call was scheduled, not that it took place. On 5 August Bishop wrote: "Attached is the doc you requested, it's the 'use of funds' spreadsheet for the designer baby and human cloning company. This gets us out of our self-funded 'garage biology' phase to the first live birth of a human designer baby, and possibly a human clone, within 5 years". The budget was "~$1.7m/year for up to 5 years + $1m for lab setup. (~$9.5m to fund 5 years, $6m for 3 years)". A human clone is a genetic copy of an existing person. Epstein replied, "no rush" (EFTA01003966, OC). On 30 August, after Bishop resent the deck, he wrote: "i like implant embryo , wait 9 months. great ending" (EFTA02625511, OC) (the page shows an encoding artefact, "=C2", after "months.").

Legality (16–17 October). Bishop reported that Japan was allowing human embryo editing in the laboratory and that a Tokyo investor had made him an offer. Epstein wrote: "we need to get a read on legal. first. cant do anything where US rules apply to us citizens regardless of where.? its such a great subject. what type of investors." Bishop answered with options: "medical tourism ... partnerships with overseas clinics and get commissions on referring the customers. We can just sell the additive DNA to those overseas clinics ... In the US, self-experimentation is not explicitly banned". Epstein: "i dont think you have enogh info to decide whether these partners are good or bad" (EFTA01019439, OC).

He Jiankui (26 November). On 26 November 2018 news broke that the Chinese scientist He Jiankui had produced twin girls whose embryos he had edited with CRISPR. That day Bishop sent Epstein the reports, a call by a CRISPR pioneer for a moratorium, and news of his own team's "new technique which does not involve stem cells in the testis, instead it's an embryo editing technique more similar to cloning". Epstein replied: "great" (EFTA01015485, OC). He's experiment was condemned worldwide. In December 2019 a Shenzhen court sentenced him to three years in prison and a fine of 3 million yuan for illegal medical practice, finding that he had forged ethical-review documents [S94, S95].

8.4 What is and is not shown

The thread shows an investor's appetite ("such a great subject", "great ending"), concern with secrecy on several sides (Hill's "total deniability" and counter-surveillance offer, Bishop's "absolute anonymity" and his reason for it, Epstein's "seen to lead"), a structurer's interest ("not bad at structuring") and one note of caution ("get a read on legal. first"). It does not show a payment, a contract or a company document involving Epstein. In February 2026 Bishop told the Daily Mail: "We never took funding from Epstein and I'm proud of that" [S89]. In his October email Bishop himself wrote of the Tokyo offer, "do I want to take /their/ money". MIT Technology Review's 2019 account of Bishop's project does not name Epstein [S39]. Separately, in July 2019 The New York Times reported, from interviews with scientists, that Epstein had told scientists he hoped to seed the human race with his DNA, using his New Mexico ranch [S01, S02, S03]. That reporting is outside the email record and is not verified here; no email checked for this paper describes such a plan.

8.5 Triad

• Ideals. "ubermensch could be the melding of humans"; "we can call it super intelligence"; "i like implant embryo , wait 9 months. great ending".
• Pushback, and decisions anyway. Bach told him the connectome plan had chances "quite close to nothing". Epstein's own one hesitation was legal ("get a read on legal. first"); no ethical objection from him appears in the thread. The caution in the Bishop thread came mostly from Bishop himself: on anonymity for the child, and on the Tokyo offer ("do I want to take /their/ money"). The decision Epstein made anyway was to keep the conversation going for four months, through the He Jiankui news, while saying that "the problem is only if i am seen to lead".
• Supporters, then and later. Hill's messages open the e-mail thread, and Bishop pitched and pursued Epstein ("you could be the first investor"); Bishop wrote that "Austin has been advising me on this" (EFTA01004839, OC; see J1-9). Later, Bishop said he never took Epstein's money [S89]. The Montreal Gazette, reporting that Hill, a co-founder and former chief executive of Blockstream, was copied on the designer-baby emails in 2018 (the government originals show Hill writing the opening messages, §8.3), quoted him as saying that his relationship with Epstein ended that year, that he regretted "any association" and "unequivocally" condemned Epstein's crimes, and: "I stand with the victims, and I support full transparency and accountability so that all responsible parties are brought to justice." The newspaper added: "The emails do not allege any wrongdoing by Hill or his colleagues" [S90].

What it points to. The designer-baby thread is the only place found in the archive where Epstein's talk of improving humans meets a concrete plan with a budget. His responses are short, and they are consistent: enthusiasm for the outcome, concern about being "seen to lead", and attention to legal exposure. The secrecy proposals themselves came first from Hill and Bishop. That pattern matches how he gave money elsewhere (Chapter 9). It is a sign of intent, not proof of a funded experiment.

9. Money, anonymity and control

Argument. The money trail shows a repeated choice. Epstein gave in ways that kept his name off the gift: anonymous gifts, gifts drafted for other donors to sign, and offers to match others' grants. Some of his money also went through investment funds. He also steered events he described as "easy to control without oversight". Some institutions refused him at the time, notably Harvard. Others, notably the MIT Media Lab, accepted the money on condition of secrecy and condemned the relationship later.

9.1 Harvard: a refusal, and the letters drafted for others

Martin Nowak's Program for Evolutionary Dynamics (PED) studies the mathematics of evolution and cooperation, for example how cooperative behaviour can survive among selfish individuals [S51, S52]. Epstein's $6.5 million gift created it in 2003, five years before his conviction [S06]. Harvard's 2020 review found that Harvard accepted no gifts from Epstein after his 2008 conviction. President Drew Faust decided in 2008 not to accept them. In 2013, when several faculty members asked Harvard to reconsider, the dean of the Faculty of Arts and Sciences, Michael Smith, again said no [S06]. This is the clearest institutional refusal in the story, and it came at the time, not after 2019.

The emails show how Epstein worked around it. On 30 July 2010 he wrote to Henry Jarecki, a psychiatrist and financier: "Henry, I thought a letter like this could accompany your contribution towards nowaks work. He is the most cited scholar in the field , no one close". Below came a ready-made letter to Harvard's recording secretary: "I would like to donate a gift in the amount of $500,000 . in support of Professor Martin Nowak's Program for Evolutionary Dynamics" (EFTA01815554, OC). On 3 December 2014 a similar letter for "$ 3,000,000 in support of Professor Martin Nowak's Program for Evolutionary Dynamics (PED)" sits in his own account, sent to himself, headed "this would mean 1 million this year and 2million anytime you wanted next year" (EFTA01001693, OC). The email does not show whose name was meant to go on it. Harvard's review found that, after 2008, Epstein "did continue efforts to interest other donors", and that in one case "Epstein or his associates helped to facilitate the gift transactions". Those donors told the review that the gifts were their own decisions and their own money [S06]. The review also found that Epstein visited the PED offices more than 40 times between 2010 and 2018 and used them to meet prominent faculty. In 2021 Harvard sanctioned Nowak over aspects of the relationship [S07, S08, S54].

9.2 MIT Media Lab: "anonymous unrestricted"

MIT took a different course. On 17 November 2013 Joi Ito, then director of the Media Lab, asked Epstein's accountant Richard Kahn to "send the $150K / year" to MIT's recording secretary: "This will be an anonymous unrestricted gift to the MIT Media Lab" (EFTA01132549, OC). An unrestricted gift is one the recipient may spend as it chooses. On 4 November 2015 Marissa Wozniak sent Ito "a draft gift letter for the anonymous donation to support Neri's research". Someone whose name is redacted ([R-7]) forwarded it to Epstein with one line: "Is anonymous from your foundation ok?" (EFTA00634804, OC; see J1-10). The attached draft read: "I hereby inform you of an unrestricted gift of $100,000 from the Jeffrey Epstein VI Foundation to support Professor Neri Oxman's research. Please note that the Jeffrey Epstein VI Foundation wishes to remain strictly anonymous with this gift" (EFTA00634805, TP).

MIT's 2020 report, by the law firm Goodwin Procter, found that Epstein made ten donations to MIT totalling $850,000. Nine of them, totalling $750,000, came after his conviction: $525,000 to the Media Lab and $225,000 to Seth Lloyd. Of the Media Lab money, $300,000 supported Joscha Bach and $125,000, given in 2015 and 2017, supported Neri Oxman. Three MIT vice-presidents knew of his conviction in 2013 and allowed the gifts on condition that they stay small and unpublicised [S04, S05]. Oxman said in 2019 that MIT required the gifts to her lab to be kept confidential "so as to not enhance his reputation by association with MIT" [S88]. In February 2019 the Media Lab rejected a further $25,000 that Epstein offered [S05].

9.3 "Easy to control without oversight"

On 1 May 2016, after Larry Summers suggested that Harvard might host an event on AI, Epstein replied: "Im happy to anonymously fund any fun gathering . There are many AI conferences , but most too large. Media Lab is easy to control without oversight and otherwise dictated polical correctness of make up of panels etc." Summers answered: "Let me ponder" (EFTA00665789, OC). The sentence is loosely typed. Its plain reading is that the Media Lab would let him shape an event, including who sat on panels, without the oversight he expected elsewhere. The same preference appears outside science. In October 2014 Summers reported telling a third party, about a gift for a poetry and video project, that Epstein "didn't do anonymous and thought best not in your name but had friend" (EFTA02361534, OC). That is, Epstein would not give anonymously, preferred that the gift not carry his name, and would give through a friend instead. This sits oddly beside his offer to Summers to "anonymously fund" an AI gathering. The common thread is not anonymity as such but keeping his own name off the gift.

9.4 Templeton matching and investment funds

Templeton. A matching gift is one paid on condition that someone else gives the same amount. In May 2009 Barnaby Marsh wrote that Templeton had "our $11M" for Nowak's programme and that raising "$6M to add" would be ideal, and that he would ask Templeton "to approve an additional $2M" (EFTA02442393, OC). In June 2009 the physicist Lawrence Krauss, director of Arizona State University's Origins Project, reported meeting Marsh. "We decided would be best to do anonymous gift, which templeton could not then promote.. he wants to start with $2M and then go to $10M if we perform ... Would you be willing to match?" (EFTA01829516, OC). Matching came up again in 2014. Marsh: "the positioning is very good for matching after we get clarity on the projects" (EFTA00987733, OC). A writer whose name is redacted ([R-10]), in a note about joining the Media Lab's advisory council: "I just jumped in to help a little with the Templeton grant, and I understand you are likely to match it" (EFTA00994276, OC). Ito: "matching grant is the key" (EFTA01805382, OC). In November 2014 Marsh described Nowak's second Templeton grant, "not yet reviewed or approved", and an offer to match (EFTA02595355, OC). No document was found showing that Epstein paid a Templeton match. Every email was searched for "templeton" or "barnaby" within 300 characters of "match", "$2M" or "10M", and the 16 distinct hits were read (gaps_phase2.md).

Investment funds. Austin Hill's 2018 remark, "I know you've hid you're investments through Joi in the past" (EFTA01004727, OC; Chapter 8), can now be set against two public statements and a summary chart of bank records. In February 2026 Hill confirmed to the Montreal Gazette that Epstein had indirectly invested US$500,000 in Blockstream through a venture fund. The newspaper reports that the fund was managed by Joi Ito, and Hill said the stake was divested in early 2015 [S90]. Blockstream's current chief executive, Adam Back, said in February 2026 that the company has "no direct or indirect financial connection with Jeffrey Epstein, or his estate", a statement about the present company that does not dispute the earlier, divested stake [S90, S97]. Hill also told the newspaper that, because the investment came through Ito's fund, "no separate vetting was conducted on (Epstein) beyond relying on professional introductions and his reputation on Wall Street", that it was less than three per cent of Blockstream's capitalization, and that it was divested "because of an unrelated conflict of interest between MIT, Joi Ito and Blockstream" [S90].

A record in the government release bears on this. On 13 August 2019 (01:13 UTC, which was the evening of 12 August, US Eastern) Justin Alfano e-mailed recipients whose names are redacted ([R-12]; the salutation is also boxed, [R-13]), with copies to Andrew Stemmer, Erika Engelson and Vijay Dewan, under the subject "Deutsche Bank Production - Transaction Charts". A document production is the set of records a party hands over in a legal process, and "Bates cites" are the page-stamp numbers given to those records. The e-mail thanks the recipients for a call "regarding Deutsche Bank's most recent document production" and attaches "several charts ... containing details and Bates cites for the transactions we highlighted on the call", under the file name "Epstein_Account_Charts" (EFTA00080250, OC; e-mail header: see J1-8). No title is printed for the sender or the recipients; the Bates prefix "DB-SDNY" and the document's place in the DOJ release suggest the charts were prepared for federal prosecutors. The charts are a summary compiled from Deutsche Bank's records, not bank statements, and the footnotes are the compiler's transcription of the "wire details". Exhibit B, "Transactions involving hedge / investment funds" (PDF p. 2, Bates EFTA00080251; footnotes on p. 10, EFTA00080259), lists payments from "Southern Financial LLC", one of the entities in the charts (an LLC, or limited liability company, is a common legal form for holding money), to "Joichi Ito": $250,000 on 2 July 2014, $500,001 on 9 September 2014 and $250,000 on 12 March 2015. It also lists $1,000,000 to "Neoteny 3, LP" on 13 April 2015 (an LP, or limited partnership, is the usual legal form of an investment fund), $483,915 from "Joichi Ito" back to Southern Financial on 5 August 2015, and $30,000 from "Neoteny 3 LP" to Southern Financial on 20 April 2018. Footnotes give the wire details, and "subcr" means a subscription, that is, buying into a fund. For the September 2014 payment footnote [3] reads: "subcr to Kyara Investments III, LLC to fund Blockstream Investment". For the March 2015 payment footnote [4] reads: "subcr to Kyara Investment IV, LLC to fund OH2 Laboratories". The footnote markers were matched to their rows on the page image. Both footnotes cite DB-SDNY-0000478, while the September row itself cites DB-SDNY-0004084. Kyara entities also appear without Ito's name. On 14 May 2014 Southern Financial paid $99,999 to "Barack Ferrazzano Kirschbaum and Nagelberg Client Fund", with footnote [2]: "Southern Financial LLC investment in Kyara Investments LLC". "Kyara Investments I" paid Southern Financial $10,000 on 2 December 2016 and $11,392 on 5 September 2018. The chart does not explain these either.

The three items can be set side by side:

• Hill, 21 July 2018: "I know you've hid you're investments through Joi in the past" (EFTA01004727, OC).
• Epstein's reply: "I dont hide" (EFTA01004727, OC).
• The chart: payee "Joichi Ito", $500,001, 9 September 2014, wire details "subcr to Kyara Investments III, LLC to fund Blockstream Investment" (EFTA00080250, OC; e-mail header: see J1-8).

The chart entry is consistent with Hill's 2026 account of an indirect US$500,000 investment in 2014. It does not show whether or when money reached Blockstream, who owned or controlled Kyara Investments III or any other Kyara entity, the purpose of the July 2014 payment, or whether the $483,915 returned in August 2015 relates to the divestment Hill describes. Whether the arrangement was concealment, or ordinary investment through a fund manager as Hill describes it, the documents do not say.

9.5 Triad

• Ideals. "Im happy to anonymously fund any fun gathering"; "Media Lab is easy to control without oversight".
• Pushback, and decisions anyway. Harvard refused his gifts in 2008 and again in 2013 [S06]. He then drafted letters for others to sign (2010) and drafted a $3 million letter (2014). Brockman and Nikolic warned him about Templeton (Chapter 3), and he pursued Templeton matching anyway. The Media Lab turned down $25,000 in February 2019 [S05].
• Supporters, then and later. At the time, Ito arranged the anonymous gifts, and the person whose name is redacted ([R-7]) asked Epstein whether "anonymous from your foundation" was acceptable. Krauss brokered Templeton. Later, Ito apologised and resigned in September 2019 [S10, S55]. MIT's report criticised the decision to accept the gifts [S04]. MIT placed Seth Lloyd on leave and later disciplined him [S14, S53]. Harvard sanctioned Nowak [S07]. Krauss's own funding from Epstein was reported in 2019 [S33].

What it points to. Wherever the money can be traced, the same preference shows: give, but keep the name off it, or put another name on it. The institutions split on this. Harvard refused outright. MIT accepted the money as long as it stayed quiet. That institutional split, more than any scientist's view of his ideas, was the main real-time check on his patronage.

10. Science as rehabilitation: work release, letters and "Jeffrey Epstein AI"

Argument. Science was also a legal and reputational tool. In 2008 his jail work release was to a "Florida Science Foundation" that a prosecutor said "exists only to employ Epstein". In 2010 he sent his lawyer "a sample letter from a scientist" to support his travel requests on probation, while an associate ran a search-ranking campaign around a science website. In 2013–14 he drafted a character reference for himself that highlighted his science, and in 2014 a writer produced articles on AI in Ethiopia and on Seth Lloyd's and Martin Nowak's work, which the scientists reviewed. Some outsiders refused to help: a reputation firm, TED, and Craig Venter's office.

10.1 The Florida Science Foundation

Work release is a jail programme that lets an inmate leave during the day to go to a job. In July 2008 an Assistant U.S. Attorney, that is, a federal prosecutor, whose name is redacted ([R-14]) wrote to a colonel whose name is also redacted ([R-15]) and to whom the decision was left ("the decision is left entirely within your discretion"). The letter opens "Dear Colonel [R-15]" and says that "During the change of plea, Mr. Epstein stated that he would be working at the Florida Science Foundation ... that the Foundation had been in existence for a 'couple of years,' and that he had been working there 'every day' prior to the plea. The Division of Corporations' documents show that the Florida Science Foundation was incorporated in November 2007, not a 'couple of years ago.'". It added that the address given for the foundation's "office" was that of his defence lawyer, Jack Goldberger (EFTA01657324, OC). Prosecutors' own compiled file records the same incorporation date (EFTA00193199, OC; see J1-7). On 20 November 2008 another Assistant U.S. Attorney, or possibly the same one (the name is redacted, [R-5]), wrote: "He is hanging out at his lawyer's office during the day, eating a lovely lunch, and getting driven back to the campgrounds at night. He is allegedly working for the 'Florida Science Foundation' which exists only to employ Epstein" (EFTA00211819, OC). A year later a prosecutor in the same office, whose name is also redacted ([R-27]), noted that "he specifically told the judge that he would be working at the Florida Science Foundation during his home confinement" (EFTA00212653, OC). Home confinement is a sentence served at home under supervision instead of in jail. On 24 December 2009 Epstein used the title himself: "I am the principal of the Florida Science Foundation. I am funding a project on quantum computing at mit, and biological code breaking" (EFTA01822876, OC).

10.2 "A sample letter from a scientist"

Probation is supervised release whose conditions limit where a person may travel. On 18 February 2010, in an email to his lawyer (whose address is redacted, [R-9]), Epstein set out what he wanted, including unlimited business or charitable travel, and ended: "i am enclosing a sample letter from a scientist , explaining why my physical presence is needed" (EFTA00763857, OC). The government PDF is one page with no attachment, and no matching letter was found anywhere in the archive. Who wrote the sample letter is unknown. The archive was searched for the phrase and for letters near the names of his funded scientists, and none of the 14 hits matched (gaps_phase2.md).

10.3 2010: "the other Jeffrey Epsteins" and jeffreyepsteinscience.com

Search-engine optimisation is the craft of pushing chosen pages up Google's results. In October 2010 Al Seckel was doing it for Epstein. He asked him to "get a tape recorder and just rabbit on your thoughts" on a list of science categories, so that the recordings could be transcribed and posted as blogs "in your own voice" (EFTA00754420, OC). He reported building sites: "As we continue to add more and more content to your real site ... The google bots have not gone around in a while ... the rankings are still working on their last go around" (EFTA00753338, OC). He pointed Epstein to "www.jeffreyepsteinscience.com" (EFTA00753832, OC). He described hiring specialists, noting that one group "wanted way to much money, well in the six figures, and so I passed" (EFTA00754633, OC). Seckel was also arranging a conference for him that autumn: "Everything is completely set for your conference" (EFTA00753978, OC). An archived press post on the science site announced that Epstein would host the "Mindshift" conference [S72].

10.4 A character reference written by himself

Between August 2013 and January 2014 Epstein's account holds about thirty drafts of a third-person reference letter about Epstein himself. The first, on 22 August 2013, reads in part: "without a doubt , one of the smartest people i have ever met, he is multilingual in the areas of advanced science. He can conduct a seminar in fields as diverse as evolutionary biology, computational neuro science ... His interests are so far ranging that encompass , signal intelligence, power law distributions , and encryption. I am well aware that in his past he has made some awful decisions, and would have done differently if he had the opportunity to do them over. however he has paid his debt to society" (EFTA01757021, OC). The To line is blank, and the drafts do not show who was meant to sign or receive the letter. The science fields named are the ones from his funding lists (Chapters 3 and 4). In this letter, science serves as character evidence.

10.5 2014: the article programme and "Jeffrey Epstein AI"

In 2014 and 2015 a writer ran Epstein's websites and placed articles. The writer's name is redacted in the 2015 report ([R-11]), and the sender line is empty or redacted in the 2014 emails. That these are the same person is suggested by the content, not shown. The programme reached the scientists he funded:

• 14 May 2014. "Could you boost the press article below?" The article was a PR Newswire release headed "The Jeffrey Epstein VI Foundation Helps Launch Artificial Intelligence in Ethiopia" (EFTA02583384, OC). The release was published on the wire [S87].
• 11 June 2014. "Let's pin down an article topic for Jeffrey by tomorrow. Can I do AI in Ethiopia? Or else I can look at what MIT 's AI lab is working on since Jeffrey funds that from time to time" (EFTA01918797, OC).
• 23 July 2014. "I am midst article on Seth Lloyd's work ... He will need to review the article as well-- just as Martin Nowak did. In the meantime, I have another article ready: AI in Ethiopia. I'm waiting for Ben Goertzel to add his input" (EFTA01853459, OC).
• 11 August 2014. "You now have 3 abstracts", listed as "Seth Lloyd-- quantum computing", "AI in Ethiopia" and "Nowak and tumor spatial analysis". "Let me know if we get traction from a publisher" (EFTA01980999, OC).
• 15 August 2014. Writing to the person preparing the article (name redacted, [R-16]), Goertzel clarified the funding: "technically, Epstein Foundation hasn't actually funded iCog Labs or the OpenCog researchers working there. Epstein Foundation has funded the OpenCog Hong Kong effort ... iCog Labs was seed-funded by Maryland investor Sander Olsen". He added that he did not mind "there being a press release associating Epstein Foundation with iCog Labs". The writer thanked him "for clarifying the funding" (EFTA00693914, OC).
• 14 June 2015. "Both your blog and main website get approx. 1.8 million viewers each per annum ... I will send you the AI in Ethiopia article tomorrow. Ben Goertzel is going to fact check it before I send it to Newsweek" (EFTA01803291, OC).

An undated draft of website copy, headed "JEFFREY EPTEIN.NET TITLE: Jeffrey Epstein AI Advancing Artificial Intelligence", instructs: "Replace image with an AI related image ... Remove Jeffrey's photo from front page ... Currently in AI: Jeffrey Epstein Helps Launch AI in Ethiopia" (EFTA00584307, TP). A draft article is titled "A Harvard Financier, Jeffrey Epstein, Advances Artificial Intelligence in Ethiopia" (EFTA01139627, TP). Neither draft shows its author. Their content matches the May–August 2014 emails, which suggests they belong to that period but does not prove it.

10.6 Refusals

• A reputation firm (October 2010). Approached through Seckel, Gregory Markel of Infuse Creative replied: "We have no problem helping someone who is innocent of accusations or a true victim of circumstance, but if there is truth to these allegations and the conviction, I'm afraid we'd have to pass" (EFTA02414937, OC).
• Craig Venter (October 2010). The genome scientist's office, through Lisa M. Betting, wrote that "he must decline your invitation to the Virgin Islands" (EFTA02421702, OC). Epstein told Krauss he was "very disappointed re venter that i have repeatedly asked you to talk to him personally . because i doubted he was actually coming" and that "so far frankly it is not very intersting at all" (EFTA01810977, OC).
• TED (January 2012). Wendy Woska wrote to Boris Nikolic that "TED has declined our transfer request because they do not feel Dr. Epstein is a good fit for the conference" (EFTA01844561, OC).

10.7 Triad

• Ideals. "I am the principal of the Florida Science Foundation"; "put it out on the blog"; the self-written reference.
• Pushback, and decisions anyway. A prosecutor wrote at the time that the foundation "exists only to employ Epstein". A reputation firm, TED and Venter's office declined. He went ahead with the work release, the probation letter, the search-ranking campaign and the article programme.
• Supporters, then and later. Nowak reviewed an article, and Lloyd was asked to. Goertzel clarified the funding for the article but did not object to the association. In 2019 Lloyd apologised to Epstein's victims [S74], and in 2020 MIT's provost announced disciplinary action against him [S53]. NPR in 2026 described philanthropy of this kind as "a significant form of power" [S59].

What it points to. Science gave Epstein a legal cover story in 2008, character evidence in 2010 and 2013, and a public identity from 2010 to 2015. The pattern of this chapter supports the reputation-repair reading most strongly.

11. How the three sides answered: then and later

Argument. Across the chapters, pushback at the time came mostly from outside Epstein's inner circle: a prosecutor, a reputation firm, TED, Craig Venter's office, Boris Nikolic on Templeton, and Harvard's leadership. Inside the circle it came as technical disagreement (Chomsky, Bach) rather than moral refusal. Moral statements arrived mostly after 2019, in institutional reports and personal statements.

Figure 3 shows who sat at the centre of the topical correspondence. By distinct topical statements, the largest nodes are Lawrence Krauss (538), Deepak Chopra (299), Noam Chomsky (283), Joscha Bach (187), Martin Nowak (187) and Joi Ito (182). Nowak is the hub that connects people: he introduced Lukin, and his office hosted the October 2017 day of meetings.

11.1 Then and later

Table 2. What participants said then and later (OC = government original confirmed; SEC = outside source)

"No public response found" means that web searches on 10–11 October 2026, covering general news and the main file-aggregator sites, turned up nothing. It does not mean that no statement exists.

11.2 Silence as an answer

The silence runs in both directions. In the emails, almost no scientist answers Epstein's grand ideas, the encrypted cell, the "female turing machine" or the "ubermensch", with a direct objection. They change the subject, ask for clarification or move on to logistics. Chomsky's polite "I don't see how to work it out" is the exception that shows how rare plain disagreement was. Afterwards, several of the people most involved in the quantum and genome strands have said nothing public that this research could find.

11.3 What changed

What changed between then and later was not the information available. Epstein's conviction was public from 2008, and Harvard, TED and a reputation firm acted on it at the time. What changed in 2019 was the cost of association. The MIT report's finding that officials allowed gifts as long as they stayed "relatively small and unpublicized" [S04, S05] describes the bargain most clearly: the scientists' work received the money, and the secrecy was meant to keep it from improving his reputation. The 2014 article programme (Chapter 10) shows that he was using that work for his reputation anyway.

11.4 How people around him responded to the allegations

In January 2015, as noted in §5.4, Kahn told Goertzel that the funding was "on hold" because of "all the bad press Jeffrey is receiving". Goertzel answered Kahn: "Sure, that is totally understood. I have seen the spate of utterly idiotic negative publicity in the news, and I'm sorry you guys have to deal with that. From my view I see two possibilities: a) that crap is all made up by greedy people seeking money and fame, b) maybe some variation of what is alleged did happen, but if so it was surely an occurrence among reasonably mature people who mutually consented at the time, so why is it anybody else's business? Our society is pretty f**ked in some ways.... But obviously you guys have already figured that out ;-p" (EFTA02373775, OC; see J1-12). The e-mail does not say which news it means. The public court docket shows that on 30 December 2014, in the federal case Doe v. United States (S.D. Fla., No. 9:08-cv-80736), two women identified in court by pseudonyms moved to join the case, and that the court denied the motion in April 2015 [S99].

This was one person's private speculation, written to Epstein's accountant while asking for money, and the public court record contradicts it. In 2008 Epstein pleaded guilty in Florida to procuring a person under 18 for prostitution. In 2011 a New York appeals court, reviewing his risk level under the state's Sex Offender Registration Act (SORA, the law that sets how closely a convicted sex offender is registered and monitored), wrote that "The evidence before the SORA hearing court established that defendant committed multiple offenses against a series of underage girls" [S98]. In February 2026 Goertzel quoted these words from his own e-mail and wrote: "I was still naively convinced of the political framing narrative, as I simply didn't question that he was lying" [S65]. To Business Insider he said that Epstein "framed it as a politically motivated prosecution for involvement with a consenting adult. I believed him. I should not have" [S64].

12. Competing explanations, limits and conclusion

12.1 Four explanations

(i) Sincere intellectual curiosity. Support: the same ideas recur for a decade in private notes he sent to himself (Chapters 3 and 4). He argued with Chomsky and Bach rather than only flattering them. Some of his questions, such as whether word meaning could be modelled as a "field", touch real research topics. Against: he rarely followed an idea into sustained work. He admitted "not very much to show for it" after 25 years of funding (EFTA00856152, OC), and his interest in deception extended to a "female turing machine" and a robot whose "age body type" he wanted to choose. Verdict: real but limited. Curiosity explains the content of the portfolio better than its form.

(ii) Reputation repair, meaning buying a respectable public image. Support: the Florida Science Foundation as a work-release employer; the "sample letter from a scientist"; the search-ranking campaign; the self-written reference; the 2014 articles and press release; and his own list of fields offered to a reporter "might better portray his intellectual philanthropic interests" (§1.1). Against: much of the giving was anonymous, which does little for a reputation. Verdict: strongly supported, and the anonymity is explained by (iii).

(iii) Access and leverage over powerful people. Support: more than 40 visits to Harvard's PED offices to meet faculty [S06]; the Media Lab as "easy to control without oversight"; gifts structured through others; scientists used as introductions to wealthy donors [S04]; and investment through funds, as with the 2014 chart entry for a payment to "Joichi Ito" described as a subscription "to fund Blockstream Investment" (EFTA00080250, OC; e-mail header: see J1-8), which the chart does not show to be concealment. Anonymity bought access without the publicity that institutions feared. Verdict: strongly supported.

(iv) An ideological project of human enhancement and control. Support: the "ubermensch" email; the "make blacks smarter" line; the designer-baby thread and the "great ending"; the racial and sexual framing of intelligence; the 2019 reporting about seeding the human race [S01]. Against: no funded enhancement project has been found. The designer-baby money never moved, by Bishop's account and in the record. His statements on these topics are fragments in emails, not a programme. Verdict: present as talk and appetite, unproven as action.

12.2 Limits

• Coverage. The email record is thin before 2009. Calendar, flight, video and audio apps could not be reached directly. Bank records appear only where a document mentions them.
• Precision. Keyword precision for the topic families is 72%, and for the three-sides searches it is very low (0–30%). The triads therefore rest on hand-selected, verified items, not on counts.
• Redactions. Twenty-six numbered redactions affect cited items ([R-3] to [R-28]; R-1 and R-2 are retired), and contact-detail boxes are counted in Appendix B. None was filled in.
• Jmail differences. Nine rows in Table J1 show places where Jmail's display differs from the government original, all of them on cited material, including two where Jmail shows names in fields the original redacts (J1-7, J1-12) or splits a message differently (J1-6).
• Absence is not proof. "No payment found" and "no public response found" describe what these searches turned up, stated with their coverage. They do not establish that nothing exists.

12.3 Conclusion

What does Epstein's funding imply about his interest in intelligence and consciousness? It implies a patron with one persistent theory and many purposes. The theory was that life is a code defended by deception, that brains and cells talk in hidden signals, and that whoever breaks the code understands the mind. He pursued it for years, mostly in emails that went unanswered and in plans to recruit code-breakers that went nowhere. The purposes were more practical. His science gave him a legal cover story, a public identity, access to elite institutions and a seat beside some of the world's best scientists. He preferred to pay without his name attached; some money went through investment funds, as a bank-chart entry for a Blockstream subscription describes; and, in the one concrete case of human enhancement in the record, insisted on not being "seen to lead".

The people around him rarely challenged the ideas, and institutions split on the money. Harvard refused it. MIT took it on condition that it stay unpublicised. Most of the moral reckoning came after his 2019 arrest and death, and some of the scientists most involved have still said nothing public. The record does not show a man who built or nearly built a new kind of intelligence. It shows how far money, curiosity and secrecy could carry a convicted man into the centre of the science of the mind, and how seldom anyone inside it said no.

Appendices

Appendix A. Jmail-only text (ST): iMessage extracts

Read this first. Everything in this appendix is ST (site text): Jmail-only text with no government original. It comes from Jmail's iMessage collection (raw/v1/imessage_messages.parquet), which the Department of Justice has not published as PDFs. The collection labels one side of each conversation "me", meaning the phone's owner, presumed to be Epstein, and the other side "them", the person the conversation is named after. Neither label could be checked against an original. None of these extracts is used as evidence in the main text. They are listed because they repeat, in Epstein's informal voice, themes that the verified emails establish.

Times are as stored in Jmail. The stored values carry no time zone, so they are not converted.

Appendix B. Verification table

Every government document cited in this paper (117 documents) is listed here, with the label it carries in the text. "Phrase checked" is the passage searched for in the PDF's text layer; author and message order were read by hand. Page images were viewed where the text layer was unclear or a redaction mattered. SHA-256 is the first 12 characters of the hash of the PDF as downloaded (blank where it was not recorded). "Boxes" is an approximate machine count of all redaction boxes on the first three pages of the PDF, numbered ones included; boxes that touch may be counted as one, and dark table-header bars may be counted as boxes (Legend, contact boxes).

Appendix C. Table J1: Jmail vs. government original

Each row shows what Jmail displays and what the justice.gov original shows. Both are shown; neither is treated as more authoritative. Rows were found in three ways: by hand, while each cited document was read against its PDF (Appendix B); by an automatic comparison of Jmail's first record with the page-1 From, To, Cc and Date lines of every cited document (file grokbot_hdr_cmp.csv, 108 documents; differences that are only formatting, such as display names, address brackets, optical-reading noise or a quoted inner header, were set aside after the flagged rows were opened on the page image); and by reading every Jmail record of the House Oversight chains cited. Names that Jmail shows in fields the original redacts are given only as Jmail's display, unverified, and are not used anywhere else in the paper. Rows J1-1, J1-3 and J1-5 of the previous version are retired because the documents they described (a 2004 flight log and a Jmail topic page) are no longer cited; the other row numbers are kept.

Appendix D. Money flows and offers named in the record

This table lists amounts that the documents name. Most are offers, requests or plans. "Status in text" says what the document itself shows; a listed amount is not proof of payment. Institutional totals from MIT's and Harvard's reports are given in Chapter 9.

Appendix E. Glossary (for reference only)

Every term below is explained in the text where it first appears. This list repeats those explanations.

• AGI (artificial general intelligence): a hypothetical machine able to learn and reason across many different tasks as well as a person, as opposed to software built for one narrow job.
• Anonymous unrestricted gift: a donation whose donor is not named publicly and which the recipient may spend as it chooses.
• Cell signalling: the chemical messages cells send to coordinate growth, repair and defence.
• Connectome: a complete wiring map of the connections between nerve cells in a brain.
• CRISPR: a molecular tool, adapted from bacteria, that cuts DNA at a chosen point so that genes can be changed.
• Cryonics: freezing a body or brain after death in the hope that future technology can revive it.
• Cryptanalysis: the craft of breaking codes and ciphers.
• Decoherence: the process by which a fragile quantum state is disturbed by its surroundings and starts behaving like an ordinary, classical object.
• Embryo (germline) editing: changing the genes of an embryo, so that the change is in every cell of the resulting person and can be passed on to their children.
• Evolutionary dynamics: the mathematics of how strategies and genes spread in populations over time; Martin Nowak's field.
• Document production: the set of records a party hands over in a legal process, page-stamped with "Bates" numbers.
• Evolutionary game theory: the mathematics of how competing strategies fare as populations evolve.
• Game theory: the mathematics of strategic choices among players whose outcomes depend on one another.
• Genome sequencing: reading out the full sequence of a person's DNA.
• Hedge fund / investment trust / offshore fund: a privately run investment pool / a company that holds investments for its shareholders / a fund registered outside the investor's home country.
• Home confinement: a sentence served at home under supervision instead of in jail.
• LLC (limited liability company) / LP (limited partnership): common legal forms for holding money and running investment funds.
• Mathematical biology: the use of equations to describe living systems.
• Matching grant: a gift paid on condition that someone else gives the same amount.
• Membrane computing: computation modelled on the way living cells pass substances across their membranes.
• Microtubules: tiny protein tubes inside cells, including nerve cells, that act as scaffolding.
• Mind uploading: the idea of copying a mind into a computer by scanning the brain.
• Orch-OR (orchestrated objective reduction): Roger Penrose and Stuart Hameroff's theory that consciousness arises from quantum events in microtubules.
• Quantum computing: computing with quantum states rather than ordinary on/off bits.
• Pulsatile signalling: signals that cells send in rhythmic pulses rather than as a steady level.
• Search-engine optimisation (SEO): shaping web pages and links so that chosen pages rank higher in search results.
• Signal processing: the engineering of pulling information out of signals such as sound, radio or sensor readings.
• Signal intelligence: the intelligence agencies' craft of intercepting and analysing other people's communications.
• Computational substrate: the physical material a computation runs on, such as silicon chips or neurons.
• SORA (Sex Offender Registration Act): the New York law that sets how closely a convicted sex offender is registered and monitored.
• Subscription (to a fund): buying into an investment fund.
• Subsystem: one working part of a larger machine that can be built and tested on its own.
• Superintelligence: an intellect far beyond the best human minds.
• Übermensch ("overman"): Friedrich Nietzsche's ideal of a human type that creates its own values; often misused to mean a biologically superior race.
• Assistant U.S. Attorney (AUSA): a federal prosecutor in a U.S. Attorney's office.
• Work release: a jail programme that lets an inmate leave custody during the day to work.

References

Outside sources (SEC). Every URL was opened on 10 or 11 October 2026 (CT); some publisher sites block automated requests, and those pages were confirmed through search-index copies.

• [S01] The New York Times (James B. Stewart, Matthew Goldstein, Jessica Silver-Greenberg) (2019-07-31). "Jeffrey Epstein Hoped to Seed Human Race With His DNA". https://www.nytimes.com/2019/07/31/business/jeffrey-epstein-eugenics.html
• [S02] The Seattle Times (NYT syndication) (2019-07-31). "Jeffrey Epstein, surrounded by scientific luminaries, hoped to seed human race with his DNA". https://www.seattletimes.com/nation-world/jeffrey-epstein-surrounded-by-scientific-luminaries-hoped-to-seed-human-race-with-his-dna/
• [S03] The Guardian (2019-08-01). "Epstein reportedly hoped to develop super-race of humans with his DNA". https://www.theguardian.com/us-news/2019/aug/01/jeffrey-epstein-seed-human-race-report
• [S04] Goodwin Procter for MIT (2020-01-10). "Report Concerning Jeffrey Epstein's Interactions with the Massachusetts Institute of Technology". https://web.mit.edu/fact2020/files/MIT-report.pdf
• [S05] MIT News (2020-01-10). "MIT releases results of fact-finding on engagements with Jeffrey Epstein". https://news.mit.edu/2020/mit-releases-results-fact-finding-report-jeffrey-epstein-0110
• [S06] Harvard University (President Bacow) (2020-05-01). "Report Regarding Jeffrey Epstein's Connections to Harvard". https://www.harvard.edu/president/news-and-statements-by-president-bacow/2020/report-regarding-jeffrey-epstein-s-connections-to-harvard/
• [S07] The Harvard Crimson (2021-03-26). "Gay Levies Sanctions Against Prof. Nowak for Contacts With Jeffrey Epstein". https://www.thecrimson.com/article/2021/3/26/gay-sanctions-nowak-epstein/
• [S08] Harvard Magazine (2021-03-25). "Martin Nowak Sanctioned for Jeffrey Epstein Involvement". https://www.harvardmagazine.com/2021/03/martin-nowak-sanctioned-for-jeffrey-epstein-involvement
• [S10] MIT Media Lab (archived) (2019-08-15). "My apology regarding Jeffrey Epstein". https://www.media.mit.edu/posts/my-apology-regarding-jeffrey-epstein/
• [S14] CNBC (2020-01-10). "MIT puts professor on leave for Jeffrey Epstein donations". https://www.cnbc.com/2020/01/10/mit-puts-professor-on-leave-for-jeffrey-epstein-donations.html
• [S16] Business Insider (2023-04-30). "Chomsky Tells WSJ His Meetings With Jeffrey Epstein Are 'None of Your Business'". https://www.businessinsider.com/noam-chomsky-mit-wsj-wall-street-journal-jeffrey-epstein-2023-4
• [S17] The Wall Street Journal (2023-05-17). "Jeffrey Epstein Moved $270,000 for Noam Chomsky and Paid $150,000 to Leon Botstein". https://www.wsj.com/us-news/jeffrey-epstein-noam-chomsky-leon-botstein-bard-ce5beb9d
• [S18] The Boston Globe (Kate Selig) (2025-11-21). "MIT's Joscha Bach: Controversial Epstein emails revealed". https://www.bostonglobe.com/2025/11/21/metro/epstein-emails-mit-joscha-bach/
• [S19] Joscha Bach (Substack) (2025-11-28). "On the Jeffrey Epstein Affair". https://joscha.substack.com/p/on-the-jeffrey-epstein-affair
• [S21] Stanford Encyclopedia of Philosophy (rev.). "Friedrich Nietzsche". https://plato.stanford.edu/entries/nietzsche/
• [S32] BuzzFeed News (Peter Aldhous) (2019-09-26). "How Jeffrey Epstein Bought His Way Into An Exclusive Intellectual Boys Club". https://www.buzzfeednews.com/article/peteraldhous/jeffrey-epstein-john-brockman-edge-foundation
• [S33] The Arizona State Press (2019-07). "Former ASU professor Lawrence Krauss received $250,000 from Jeffrey Epstein". https://www.statepress.com/article/2019/07/former-asu-professor-lawrence-krauss-received-250000-from-jeffrey-epstein
• [S35] Progress in Biophysics and Molecular Biology 190:28-169 (2024). "A landscape of consciousness: Toward a taxonomy of explanations and implications". https://pubmed.ncbi.nlm.nih.gov/38281544/
• [S36] Physics of Life Reviews 11(1):39-78 (2014). "Consciousness in the universe: A review of the 'Orch OR' theory". https://pubmed.ncbi.nlm.nih.gov/24070914/
• [S39] MIT Technology Review (Antonio Regalado) (2019-02-01). "The DIY designer baby project funded with Bitcoin". https://www.technologyreview.com/2019/02/01/239624/the-transhumanist-diy-designer-baby-funded-with-bitcoin/
• [S40] AI Magazine 25(2) (2004). "The St. Thomas Common Sense Symposium: Designing Architectures for Human-Level Intelligence". https://cogaffarchive.org/AIMag/singh-minsky-sloman-aimag-04.pdf
• [S41] WIRED (Steven Levy) (2025-08-01). "Inside Jeffrey Epstein's Forgotten AI Summit". https://www.wired.com/story/epstein-files-science-tech-funding/
• [S43] Business Insider (2019-09). "A Journalist Claims Epstein Told Him He Funded a Robot Named Sophia". https://www.businessinsider.com/jeffrey-epstein-told-a-journalist-he-funded-sophia-the-robot-2019-9
• [S45] American Journal of Human Genetics 103(5) (2018). "ASHG Denounces Attempts to Link Genetics and Racial Supremacy". https://pubmed.ncbi.nlm.nih.gov/30348456/
• [S46] The Harvard Crimson (2025-11-18). "Summers To Step Back From Public Commitments Amid Epstein Scandal". https://www.thecrimson.com/article/2025/11/18/summers-steps-back-from-public-commitments-epstein/
• [S47] NBC News (2025-11). "'I am deeply ashamed': Larry Summers to step back from public commitments after new Epstein emails". https://www.nbcnews.com/politics/politics-news/-deeply-ashamed-larry-summers-jeffrey-epstein-emails-rcna244526
• [S49] STAT (archived) (2019-08-05). "Citing 'nerd tunnel vision,' biologist George Church apologizes for contacts with Jeffrey Epstein". https://web.archive.org/web/20190805205504/https:/www.statnews.com/2019/08/05/citing-nerd-tunnel-vision-biologist-george-church-apologizes-for-contacts-with-jeffrey-epstein/
• [S50] BuzzFeed News (Peter Aldhous) (2019). "Scientists Took Jeffrey Epstein's Money After He Was Jailed For Sex Crimes. Here's What They Have To Say About It Now.". https://www.buzzfeednews.com/article/peteraldhous/jeffrey-epstein-science-donations-apologies-statements
• [S51] Science 314:1560-1563 (2006). "Five Rules for the Evolution of Cooperation". https://www.science.org/doi/10.1126/science.1133755
• [S52] Free Press / martinnowak.com (2011). "SuperCooperators: Why We Need Each Other to Succeed". http://www.martinnowak.com/books/supercooperators
• [S53] MIT Provost Martin Schmidt (2020-12-18). "Decision on Professor Seth Lloyd". https://orgchart.mit.edu/letters/decision-professor-seth-lloyd
• [S54] The Harvard Crimson (2020-05-02). "FAS Places Prof. Nowak On Leave After Report Finds Epstein Used His Program to Rehabilitate Image". https://www.thecrimson.com/article/2020/5/2/epstein-review-nowak-leave/
• [S55] NPR (2019-09-07). "Joichi Ito Steps Down As MIT Media Lab Director Over Ties To Jeffrey Epstein". https://www.npr.org/2019/09/07/758670699/m-i-t-media-lab-director-steps-down-over-financial-ties-to-jeffrey-epstein
• [S59] NPR (Katia Riddle) (2026-03-30). "Philanthropy is a 'significant form of power.' Here's how Jeffrey Epstein exploited that.". https://www.npr.org/2026/03/30/nx-s1-5753946/science-jeffrey-epstein-mit-exploit-donors-philanthropy
• [S61] USA Today (Melina Khan) (2026-02-23). "Deepak Chopra acknowledges 'poor judgment' over emails with Epstein". https://www.usatoday.com/story/entertainment/celebrities/2026/02/23/deepak-chopra-jeffrey-epstein-emails/88821384007/
• [S62] CNN (2026-02-23). "Chopra had close relationship with Epstein, told him to bring 'your girls' on trips". https://www.cnn.com/2026/02/23/politics/deepak-chopra-jeffrey-epstein-files
• [S63] South China Morning Post (2026-02-16). "Epstein funded AI pioneer, helped him get HK$8.9 million in Hong Kong government grants". https://www.scmp.com/news/hong-kong/society/article/3343645/epstein-funded-ai-pioneer-helped-him-get-hk89-million-hong-kong-government-grants
• [S64] Business Insider (Hugh Langley) (2026-02-19). "AGI Researcher Ben Goertzel Courted Epstein for Funding, Emails Show". https://www.businessinsider.com/ben-goertzel-jeffrey-epstein-agi-researcher-emails-jail-2026-2
• [S65] Ben Goertzel (Substack) (2026-02-20). "Goertzel vs Epstein". https://bengoertzel.substack.com/p/goertzel-vs-epstein
• [S66] Daily Wildcat (Jasmine Creighton) (2026). "Epstein files release outlines two UA professors' relationship with Jeffrey Epstein". https://wildcat.arizona.edu/164924/news/epstein-files-release-outlines-two-ua-professors-relationship-with-jeffrey-epstein/
• [S72] jeffreyepsteinscience.com (Epstein PR; archived) (2010-11-06). "Jeffrey Epstein to Host Mindshift Conference". https://web.archive.org/web/20110620014056/http:/www.jeffreyepsteinscience.com/2010/11/jeffrey-epstein-to-host-mindshift-conference/
• [S73] CounterPunch (2026-02-09). "Ensnared and Manipulated: Official Statement by Valeria Chomsky Regarding Jeffrey Epstein". https://www.counterpunch.org/2026/02/09/ensnared-and-manipulated-official-statement-by-valeria-chomsky-regarding-jeffrey-epstein/
• [S74] Seth Lloyd (Medium) (2019-08). "I am writing to apologize to Jeffrey Epstein's victims". https://medium.com/@sqlloyd/i-am-writing-to-apologize-to-jeffrey-epsteins-victims-eee805c4f13
• [S78] U.S. House Committee on Oversight and Government Reform (2025-11-12). "Oversight Committee Releases Additional Epstein Estate Documents". https://oversight.house.gov/release/oversight-committee-releases-additional-epstein-estate-documents/
• [S83] The Globe and Mail (2025-11-14). "Canadian physicist Lawrence Krauss sought Epstein's advice on how to respond to sex-misconduct allegations, files show". https://www.theglobeandmail.com/world/article-jeffrey-epstein-files-canadian-physicist-lawrence-krauss/
• [S84] Developmental Medicine & Child Neurology 18(5):561-567 (Charles M. Super) (1976). "Environmental Effects on Motor Development: the Case of 'African Infant Precocity'". https://doi.org/10.1111/j.1469-8749.1976.tb04202.x
• [S85] American Psychologist 67(2):130-159 (Nisbett Aronson Blair Dickens Flynn Halpern Turkheimer) (2012). "Intelligence: New Findings and Theoretical Developments". https://doi.org/10.1037/a0026699
• [S87] PR Newswire (2014-05-14). "The Jeffrey Epstein VI Foundation Helps Launch Artificial Intelligence in Ethiopia". https://www.prnewswire.com/news-releases/the-jeffrey-epstein-vi-foundation-helps-launch-artificial-intelligence-in-ethiopia-259205671.html
• [S88] Neri Oxman (via DocumentCloud / Boston Globe) (2019-09-13). "Statement from Neri Oxman". https://s3.documentcloud.org/documents/6408615/Statement-From-Neri-Oxman.pdf
• [S89] Daily Mail (2026-02). "Epstein linked to 'designer baby' empire in latest files". https://www.dailymail.com/sciencetech/article-15520573/Jeffrey-Epsteins-baby-cloning-new-files.html
• [S90] Montreal Gazette (Harry North) (2026-02-06). "The Montreal tech entrepreneur named in the Epstein files". https://montrealgazette.com/news/montreal-meeting-with-jeffrey-epstein-island-files/
• [S91] Physical Review E 61:4194 (2000). "Importance of quantum decoherence in brain processes". https://doi.org/10.1103/PhysRevE.61.4194
• [S92] Physical Review E 65:061901 (2002). "Quantum computation in brain microtubules: Decoherence and biological feasibility". https://doi.org/10.1103/PhysRevE.65.061901
• [S93] Nature 440:611 (2006). "Quantum mechanics in the brain". https://doi.org/10.1038/440611a
• [S94] Xinhua (2019-12-30). "He Jiankui gets 3 years for illegal human embryo gene-editing". https://www.xinhuanet.com/english/2019-12/30/c_138666463.htm
• [S95] Nature (news) (2020-01-03). "What CRISPR-baby prison sentences mean for research". https://doi.org/10.1038/d41586-020-00001-y
• [S96] NobelPrize.org (2020-10-06). "The Nobel Prize in Physics 2020". https://www.nobelprize.org/prizes/physics/2020/summary/
• [S97] The Block (2026-02-02). "Blockstream has no financial links to Jeffrey Epstein, Adam Back says". https://www.theblock.co/news/business/2026-02-02-blockstream-jeffrey-epstein-adam-back-387956
• [S98] New York Supreme Court, Appellate Division, First Department (official reporter, nycourts.gov; also on Justia) (2011-11-17). "People v Epstein, 2011 NY Slip Op 08293 (App. Div. 1st Dept.)". https://www.nycourts.gov/reporter/3dseries/2011/2011_08293.htm
• [S99] U.S. District Court, S.D. Fla. (via CourtListener) (2015-04-07). "Doe v. United States, No. 9:08-cv-80736 (S.D. Fla.), Order, DE 324 (entered 7 Apr 2015), describing the 30 Dec 2014 joinder motion". https://www.courtlistener.com/docket/4502088/324/doe-v-united-states/