Gino Yu

Named on the documents as	”Gino Yu” in the sender field of his own messages; “Dr. Gino Yu / Associate Professor / Multimedia Innovation Centre, School of Design / Hong Kong Polytechnic University / Hung Hom, Hong Kong” in the address block of a sponsorship letter; “Ben Goertzel and Gino Yu, Principal Investigators” on the title page of a research proposal; “gino”, lower case and without a surname, throughout Jeffrey Epstein’s own messages and his office’s day-sheets123
What a document in the release says he is	An associate professor at the Hong Kong Polytechnic University’s School of Design, principal investigator on a Hong Kong government technology grant, and joint principal investigator with Ben Goertzel of an open-source artificial-general-intelligence project proposed to Epstein’s foundation124
Range of documents cited here	A Novamente LLC sponsorship letter written before the Xiamen office it describes was due to open in late 2010, through to a message of 31 January 2019 introducing Epstein to a man he had met at Davos15
Size of the file	1,302 documents read as his. The phrase "Gino Yu" returns 1,086 of them, 83 per cent. The other 216 are reachable only through a damaged spelling or through the bare forename gino, which is how Epstein and his office almost always wrote him67
How much of it is his own writing	330 of the 1,302, 25 per cent, carry his name at the head of the top-level sender field of the document’s own first sheet, under the rule published below. 291 of those 330 are addressed to Epstein. They resolve to between 290 and 322 distinct messages, running 7 May 2015 to 31 January 2019. A further 332 address him without his writing anything8
What he did for Epstein	Introduced him to scientists, conference organisers, writers and investors; ran and staffed conferences Epstein paid for; installed a virtual-reality rig at Epstein’s New York house; and proposed building “a base at our University” where people Epstein supported “could thrive, develop new stuff, and also be studied”91011
What Epstein’s money did	Epstein offered him $100,000 in 2016, of which Yu asked that $25,000 go to a Havana conference; Epstein wrote on 15 March 2017 “i need all details on cuba i have sent 25k already”; and on 8 March 2018 he instructed his accountant, copying Yu, “rich , send gino 25k from gratitiude. forundation. so he can find specali ability people”91213
What Epstein said about him	To one correspondent on 5 November 2017: “gino says he will be with you, careful, he is a bit of a huckster”. To another on 4 January 2018: “gino is a bit too mjuch of a hustler”314
His answer in the outside record	None. Hong Kong reporting in February 2026 recorded that he had not responded to requests for comment15
Not established here	That any of the sums instructed or offered reached him; that the United States non-profit he offered as a conduit ever received anything, or was used at all; what the “developmental model” he presented to Epstein consisted of; that he knew anything of Epstein’s conduct; anything about him beyond what the sheets cited here record

Gino Yu is a Hong Kong academic who, between 2015 and 2019, was one of Jeffrey Epstein’s most frequent correspondents and his most productive supplier of other people. The file is not a file about a scientist Epstein funded. It is a working relationship with three strands. The first is procurement: Yu found people and delivered them, physicists and neuroscientists and writers and investors and people he described as having “special abilities”, and he described the job to Epstein himself in one phrase, “Happy to serve as your science pimp.” The second is money moving outward through him: a hundred thousand dollars offered in 2016, twenty-five thousand of it sent to a conference in Havana, and a further twenty-five thousand instructed out of Epstein’s foundation in March 2018. The third is a proposal, made in October 2017, that the people Epstein supported be housed and studied at Yu’s university. Nothing on this page is a finding of wrongdoing.

Finding him: the phrase is not the problem, the forename is

This subject inverts the usual trap. His full name is unusual and the exact phrase for it is almost clean. What the exact phrase cannot reach is the way everyone around Epstein actually referred to him, which was by forename alone.

The name phrase. Measured on 2026-09-12 against the corpus full-text index, which covers the EFTA, HOUSE_OVERSIGHT and DOJ-OGR productions together, and counting distinct documents: "Gino Yu" returns 1,086. Seven further documents are reached by three damaged spellings and by nothing else: "Gino Vu" (1 document), "Gino Yn" (1) and "Gino Y" (20 documents, 5 of them outside the base). Every one of the seven was opened and every one is this subject. The union of all forms is 1,093.⁶¹⁶

The variant generator produces nothing usable. ocr_variants.py returns exactly one candidate for this name, "Gin0 Yu", and that form returns zero documents. Sixteen further forms were hand-probed and all returned zero. The three damaged spellings that exist were found by reading.⁶

Never key a figure on the surname. Bare "Yu" returns 2,194 documents and this subject is a minority of them. Two things inflate it. One is other people, a “Philip Yu” on Epstein’s own schedules foremost among them. The other is the extraction, which renders the ordinary word you as yu: “do yu have”, “where are yu”, “see yu tomorrow”. 1,105 documents carry a Yu token and are not this man, and no figure on this page is keyed on that token.¹⁷

The forename is where he actually lives. Bare "Gino" returns 1,452 documents, of which 359 fall outside the union of every form of his full name. Those 359 were opened one by one and the line carrying the string was read in each. They do not reconcile into a residue; they partition, and the largest bucket by a wide margin is him:

what the 359 are	documents
this subject, under the bare forename or a damaged possessive	209
extraction noise carrying no name at all	54
Gino Severini, the painter, in Epstein’s art schedules	28
Gino Cicerchia, a Palm Beach estate agent, in social-network notifications	14
Gino Segrè, the physicist, in mailing-list address blocks	12
twelve other people and businesses whose name a sheet prints as Gino	12
Gino Vizzari, on a jail housing list	10
Gino Silvestri, of a state money-laundering task force, on subpoena returns	6
”Gino Avenue”, which is a street name	4
Francesca Gino, where Gino is the surname	4
a news URL naming James Gino Salerno	2
a cook, in two messages of March 2010	2
Gino Pereira, in an enforcement memorandum	1
Ginevra Tamberi, whom the sheet itself calls “Gino’s”	1

209 + 54 + 28 + 14 + 12 + 12 + 10 + 6 + 4 + 4 + 2 + 2 + 1 + 1 = 359.⁷

The 209 are not marginal. They are Epstein’s own instructions (“Ask Gino if we can also do another five on wed”, “rich , send gino 25k from gratitiude”, “why dont you ask gino , i think one of the conferences he runs is a charity in the states”, “gino? this morning? leon? miro ?”), his private assessments of the man, and the office’s day-sheets. The most operational traffic about this subject is precisely the traffic his full name cannot reach.⁷¹⁸

One line, more than twenty spellings, half of them invisible to the name. The Epstein office schedule for Thursday 19 November 2015 carries a reminder that “George Mueller (Gino Yu’s friend) to arrive NY”. Forty-eight copies of that reminder survive in the release, and the extraction renders the three words in more than twenty different ways. The exact number depends on how leading punctuation and a line break inside the string are normalised, and the normalisation used here is published in the footnote. What does not depend on it is that thirteen of those renderings contain no trace of his surname at all: Gino Nes, Gino lit:s, Gino lit :s, Gino lits, Gino Ms, Gino Hes, Gino Pis, Gino Pr’s, Gino Pt’s, Gino Yies, Gino Yit:s, Gino Yir:s and Gino Yith. One of those sheets was rendered locally and read at 500 dpi: it prints “Reminder: George Mueller (Gino Yu’s friend) to arrive NY,”, plainly, in italic type, followed by a black bar. The apostrophe is what defeats the extraction, and 23 of the 48 copies are reachable by no form of his name.¹⁹

A different kind of name fault sits inside this population, and it is the document’s fault, not the extraction’s. One of the 359 carries the line “Gino’s reply re ‘when next’: It will be very unlikely that it will be before Cannes in May.” The sheet was rendered and read. It prints “Gino’s”. But the message quoted directly beneath it is signed on two lines, each prefixed on the sheet by the quote marker ”>”, “Gini” and then “Ginevra Tamberi”, and concerns a film print being sent back to the laboratory before the Cannes festival, and its subject line reads “Gini in Woody’s office re Brad Karp screening”. The writer typed the wrong name. She was Lesley Groff: the sender value is barred on this sheet, but two other sheets of the same thread carry her name unbarred, one of them as its own sender value and the other in both its To: line and the quoted attribution line naming her as the writer of the forwarded message. Neither of those two contains the string “Gino” anywhere. A reconciling total cannot see this and neither can an image of the one sheet: the image confirms the error instead of correcting it. What corrects it is reading the sheet through to the signature, and then reading the copy set. That document is assigned to Ginevra Tamberi above and nothing on this page rests on it.²⁰

Taken together, the documents read as his are 1,302: the 1,093 reachable by some form of his name plus the 209 reachable only by the forename. The obvious phrase reaches 83.4 per cent of that.

How much of this file is his own writing

A count of documents naming a person is not a count of what that person did. The rule below is stated completely enough to be re-implemented, because the tool that would otherwise answer this question is wrong here by seventy-nine documents.

The rule. For each document, take its own first sheet, meaning the row with the lowest page number for that identifier, and the first twelve non-blank lines of it. Find the first line matching a sender label anchored at line start, over the damaged forms of From: the extraction produces (Fran, Front, Frorn, Frnm, Fram, Fnxn, Fromm, Fm, Frm, Fron, Erom, Frcrn, Ftom, Fiom, Prom, Sender). The sender value is the remainder of that line if it is non-empty; otherwise the next non-blank line, unless that line is itself a header label or begins with a bare header word, in which case the sender value is absent and the document is declined. Only the first sender label is used, so a quoted header lower on the sheet can never supply the sender.

The recipient rule, at the same grain, because this page also prints recipient figures and a figure without a rule is what this page criticises elsewhere. In the same first sheet and the same twelve-line window, take every line matching To:, Cc: or Bcc: anchored at line start; each such line’s value is the remainder of that line if it is non-empty, otherwise the next non-blank line, unless that line is itself a header label or begins with a bare header word, in which case that label contributes nothing. The document’s recipient set is the union of the values so taken. Unlike the sender rule this one is not first-label-only, because a message has more than one recipient. And the two name tests applied to those values, because both of them move the figure by three documents. A recipient value names him when it carries the forename gino at a word boundary or carries it run straight into a following surname where the extraction has dropped the space: three To: values in this population are rendered GinoYu<ta, GinoYu<ta and GinoYul, and a word-boundary test alone cannot see any of them. A recipient value names Epstein when it carries jeffrey E., Jeffrey Epstein or the string jeevacation however the domain renders after it: three To: values here are rendered jeevacation®gmail.com, jeevacationalmail.com and jeevacationiamaitcom.⁸

The timestamp rule. In the same sheet and window, take the first line matching Date: or Sent:, with its value read the same way, and parse it under the nine header formats this release uses. Where the value carries a UTC offset the value is converted to UTC before two documents are compared; where it carries none it is compared as printed. Because that choice is a choice, this page prints what both choices give.

The result, over the 1,302: 1,069 documents yield a top-level sender; 187 carry a sender label whose value is absent, barred or blank and are declined; 46 carry no sender label at all.

• 330 documents, 25 per cent, carry his name in that top-level sender value, and in all 330 his forename stands at the head of it.
• 291 of the 330 are addressed to Epstein, at jeffrey E., Jeffrey Epstein or [email protected]. That is 88 per cent of his own outbound traffic in this release.
• 332 documents address him in a To: or Cc: value without his name appearing in the sender value.

The figure does not move, and it is not the tool’s figure. Re-running the same rule at windows of 8, 12, 20 and 30 non-blank lines gives 330 every time. The desk’s act_rate.py, whose self-test printed PASS before any figure here was taken, returns 405 of 1,086 on the base phrase where this rule returns 326. Of the 85 documents the tool scores and this rule does not, 50 are sheets on which the tool has joined two or more sender labels together, and on 67 of the 85 the first sender label on the sheet is somebody else, usually Epstein or Lesley Groff quoting one of Yu’s messages underneath. On this subject the tool overstates by a quarter; on people/reid-hoffman the same fault made it understate. Neither number should be quoted without its rule, and this page quotes 330 and has just given the rule.⁸

330 documents are not 330 messages. Of the 330, 315 yield a parsable top-level timestamp under the rule above and 15 do not. Those 315 resolve to 290 distinct timestamps when an offset is converted to UTC before comparison, and to 307 when the values are compared exactly as each sheet prints them. With the unparsed added at worst one for one, his own outbound traffic is between 290 and 322 distinct messages, and the spread between those two figures is the cost of the normalisation choice, not a measurement error. By year, the 315 dated documents fall 16 in 2015, 125 in 2016, 130 in 2017, 41 in 2018 and 3 in 2019. His own writing collapses after 2017 and very nearly stops in the last year of the file, which is the opposite of the shape of the population as a whole.⁸

He is an academic who ran conferences, and the standard press test is right about him for once. act_rate.py reports zero press documents here. That zero was not taken on trust: all 46 documents whose first sheet carries no sender label at all were opened and read, because that is where a press capture would hide. There are none. They are 20 Epstein office schedules, 10 papers and funding proposals by Ben Goertzel, 7 device-forensics exports, 3 copies of the roboticist David Hanson’s biography, 2 conference programmes, 2 Skype conversation logs, the Novamente sponsorship letter, and one office message whose sender field is barred. Running the tool’s own press expression over the whole first sheet rather than its 1,200-character window fires on 203 of the 1,302, and on 202 of them the only marker matched is “All Rights Reserved”, which is the last line of Epstein’s own confidentiality footer. Not one of the 203 is a press document.²¹

What the paper trail says he is

The earliest document read for this page is a letter on Novamente LLC letterhead, signed by Ben Goertzel, addressed to “Dr. Gino Yu / Associate Professor / Multimedia Innovation Centre, School of Design / Hong Kong Polytechnic University / Hung Hom, Hong Kong”. It says: “We are excited to support your proposal titled ‘A Software Toolkit for Creating Intelligent Non-Player Characters in Video Games.’” It offers “HK$280,000 in cash as matching fund sponsorship, should your proposal be selected for funding”, and it refers to an office in Xiamen “that we intend to open in late 2010”.¹

What that letter was matching is set out in a research proposal in the release addressed to the Epstein Foundation and dated February 2013: the money “is already partially in place, due to an ITF (Innovation in Technology Fund) grant obtained via the Hong Kong government, with Dr. Gino Yu as Principal Investigator and Ben Goertzel’s AI consulting firm Novamente LLC as corporate sponsor (providing 10% of the funds to match the ITF’s 90%)”.⁴

A second, undated proposal in the release carries the two men as equals on its title page: “OpenCog / AGI Toddler Project / Open-Source ‘Human Toddler Level’ AGI by 2015 for US $3M Total / Ben Goertzel and Gino Yu, Principal Investigators / Multimedia Innovation Centre, P507 Hong Kong Polytechnic University”. Its management section provides that “the management board will initially be composed of Dr. Ben Goertzel, Dr. Gino Yu and one or more representatives appointed by the project donors.”²

So the route into Epstein’s file runs through a laboratory and a grant, and it predates the correspondence by years. These proposals sit in the release beside one headed “Jeffrey Epstein IV Foundation / AGI Initiative / R&D Funding Proposal / Ben Goertzel / December 2, 2013”. No document read for this page shows what, if anything, the foundation paid against any of them.⁴

2015: he is already in the house

The correspondence itself opens in May 2015 with Yu and the cognitive scientist Joscha Bach being fitted into Epstein’s day. On 7 May 2015 Lesley Groff wrote to Bach, “Good morning Joscha. Might you and Gino be available to come see Jeffrey today? Perhaps 1:30?”; Bach answered “Sure! I will ask Gino, too.”; and Groff’s reminder to Epstein that morning read “Reminder Gino Yu and Joscha were saying you wanted to see them again before you leave.”²² Yu’s own earliest message in this file is of the same day and its whole body is his name: subject “Available tomorrow!!”²³

Three days later he was already doing the thing he would do for the next four years. On 10 May 2015 he wrote, for forwarding to Epstein: “I’m seeing Dmitry Itskov who is a radical ‘out of the box’ thinker: http://2045.com/ next week. He’s been funding radical work (keeping brains alive and interfaced to computers) and relocating from Moscow to NY. Let me know if you’d be interested in having a conversation with him.”²⁴

By November 2015 the office was describing visitors by their relation to him. The schedule for Thursday 19 November 2015 opens the day with “Reminder: George Mueller (Gino Yu’s friend) to arrive NY”, two entries above a breakfast with Joi Ito and three above a lunch with the mathematician Misha Gromov: the reminder, then “8:00am Appt w/Justin Nelson”, then “8:30/9:00am BREAKFAST w/Joi Ito”, then “12:30pm LUNCH w/Misha Gromov”.¹⁹

The supply of people

The single most common shape in his own outbound mail is an introduction. Five of his messages are headed “Intros”, four “Introductions”, and dozens more carry a name in the subject line. The pattern is constant: Yu tells Epstein who someone is, tells the other person who Epstein is, and hands the arrangement to Epstein’s office.

To third parties he had a standing formula for Epstein. To a Hong Kong research publisher in December 2016: “Can I forward this to my benefactor in NYC? You interested in meeting him while you are there?”²⁵ To the son of a prospective guest in August 2016: “I’ve cc’d my benefactor, Jeffrey Epstein, to this mail. Would love for your dad to meet with him on Sept 6th.”²⁶ To a conference organiser in February 2017: “I’ve cc’d my benefactor, Jeffrey Epstein. He can make arrangements on the transfer of funds for the Conference. Jeffrey is also really well connected and can bring significant relationships to make this an amazing event.”²⁷ “My benefactor” appears in 39 documents of this population.⁶

The office wrote the other half of the same transaction. On 22 May 2018 Groff wrote to the economist Nouriel: “Hello Nouriel. I am Jeffrey Epstein’s assistant…Gino Yu suggested the two of you meet up in NY”; the schedule entry for that appointment reads “4:00pm Appt w/Nouriel (friend of Gino Yu)”. An October 2016 schedule carries “8:00pm Appt w/Stuart Hameroff (friend of Gino Yu)”. A January 2017 schedule carries “6/6:30pm Appt w/Ravi Mantha (friend of Gino’s)”. An October 2016 schedule carries “Reminder: Peter Fenwick and Ray (Gino’s people) were asking if they could come see you today”.²⁸

The people he delivered are named across the file: the near-death-experience researcher Peter Fenwick and the physicist Ray F. Lee; the anaesthesiologist Stuart Hameroff, with an offer to bring Sir Roger Penrose too; the writer Lydia Laurenson, who wrote to Epstein on 21 October 2017, “any friend of Gino’s, etc.”; and, in January 2019 after the World Economic Forum, two more men “interesting to meet.”²⁹⁵

One of those introductions is the first link of a money chain set out elsewhere on this wiki. On 17 August 2016 Yu sent Epstein a Closer To Truth episode link under the subject “Kuhn Interview… Divine intervention.” The next day Epstein wrote back to him: “can you introduce me to kuhn by skype.” On 20 August Yu wrote to Robert Lawrence Kuhn: “I’d like to introduce you to Jeffrey Epstein who I was going to see if you could meet with while I was in NYC. He has an interest in AI and consciousness and I showed him your Closer to Truth site.” Kuhn answered both men on 21 August, opening “Hi Gino: Appreciate the intro.” What followed, a thirteen-episode public-television pitch naming the Jeffrey Epstein VI Foundation, a $150,000 payment in 2017 and a $500,000 instruction in 2018, is set out on Robert Kuhn’s page and is not repeated here.³⁰

The house, the headset, and the girls

In October 2016 Yu installed hardware at Epstein’s New York house. Groff wrote to Epstein on 25 October, “What room would you like Gino to set up a VIVE at the house tomorrow? He will need to come over around 11 am or so to prepare for his 4pm meeting with you”; Epstein’s answer was “small room off library”. A schedule for the following day carries the entry “(TBD Gino Yu to arrive and set up VIVE in the small room off of the library)”. The next evening Groff wrote under the subject line “Gino is in LA! 4pm is cancelled!”, and her message began “Gino is in LA!! for some reason he thought the appointment was for next =eek!!”. The = is printed on the sheet, in the word “week”.⁹

The purchase had been proposed four days earlier and its funding named. On 21 October 2016 Yu wrote to Epstein, under the subject “Do you have a Vive?”:

I’m getting you and your girls a Vive to play with. Ideally funded out of the 100K. Just want to make sure you don’t have one yet. Will install at your place when I’m in NYC.

Epstein’s whole reply the next morning was “Donot have”. This sheet was rendered locally at 300 dpi and read.⁹

The same phrase had appeared in August. Epstein had been testing a man who claimed to be able to identify people from photographs, and wrote on 7 August 2016, “why dont we just put david to his own claim. 25 photos. 25 names. randomize and see how he does. ? without that first test I have no further interest.” Yu answered at length the same day with a protocol, and then proposed a subject:

As he’s read one of your girls (Karyna?), perhaps we could task her to get 25 pictures of her friends and people that she knows.

Two later paragraphs of the same message name the budget and then offer a job description. They are not adjacent on the sheet: a postscript about the Shanghai conference stands between them, and the second is the one the sheet opens “PPS.”

Whatever the costs to set up, you can take it out of the 100K (just leave 60K for the conference events that ███ like to run). ███ also happy to fly out to NYC to witness the experiment.

With another budget allocation, I am happy to work with you on interesting questions and bringing the best people together to discuss them (the budget would cover travel and accommodations). Happy to serve as your science pimp. After spending the day with you, I value your approach and insights as much as your funding.

That sheet was rendered locally at 500 dpi and read for this pass; the paragraph about the 100K carries two black bars, one in each of its two sentences, and both are shown above with nothing supplied in their place. Karyna Shuliak was Epstein’s girlfriend and has her own page here; no document read for this page shows that she was asked, that she agreed, or that the experiment took place.³¹

The money, as the documents have it

Three sums are named in this file and they are named by the men who were moving them.

The $100,000, offered in 2016. Its existence is established by Yu’s own accounting of it rather than by any instrument. On 7 August 2016 he wrote of “the 100K (just leave 60K for the conference events that ███ like to run)”; on 21 October he proposed the headset be “funded out of the 100K”; in February 2017 he asked Epstein “what conditions you have on the remainder of the 100K”; and in a later message he offered to “use the remainder of the 100K funds that you’ve offered me a while back.”³¹⁹³² On 25 November 2017 he set out what had come of it:

When you offered me the 100K last year, my only request has been to support the Cuba Conference when they asked for 25K. My travel to see you, introductions to others, and the Vive that I bought for you, but have not yet installed, have come out of pocket.

By his own account, then, the only part of the hundred thousand that he asked to be spent was the twenty-five thousand for the conference, and the headset he had promised a year earlier was still not installed.³³

The $25,000 to Havana, sent March 2017. The conference was an international neuroscience meeting held in La Habana on 6 to 8 November 2017, co-chaired by professors in Havana and Rome. On 30 January 2017 Yu wrote to Epstein, “For the Cuba conference, they are looking for the sponsorship (25K). Who should I direct them to?” On 21 February he told the Rome co-chair, “My benefactor approved 25K. Am working out the contact for the transfer”, and copied Epstein in. On 15 March 2017 Epstein wrote to him: “i need all details on cuba i have sent 25k already”. After the conference the co-chair wrote to Epstein directly: “No words enough to express gratitude for your enormous support, that made the event possible.”¹¹¹²³⁴

The $25,000 from the foundation, instructed March 2018. At 9.06 p.m. on 8 March 2018 Epstein wrote a single line to his accountant Richard Kahn, copying Yu:

rich , send gino 25k from gratitiude. forundation. so he can find specali ability people

That sheet was rendered locally at 300 dpi and read; both recipients’ addresses are covered by black bars. Yu answered the next afternoon, thanking Epstein and volunteering a conduit:

I have an account in Hong Kong that is commercial, but also have access to a non-profit in the US that can pass funds (http://hinmancharity.org/about/) Would the non-profit be of benefit to you for tax purposes?

Will keep all the receipts and short description of the use of proceeds for each item.

That sheet was rendered locally at 600 dpi and read for this pass; it prints the address as a live hyperlink ending in a slash, and Yu’s own address and Kahn’s are covered by black bars. That is the same question, asked from the other side, that Epstein had put to Robert Kuhn thirteen months earlier: “is it a charity that received the funding?” The pattern on both files is a payment from Gratitude America, Ltd. and a recipient offering a United States charity to receive it.¹³

No payment to him appears in the Gratitude America records set out on that entity’s page, whose cheque series is expressly not complete. The corpus documents show an instruction; they do not contain a cheque, a wire confirmation or a receipt for any of the three sums.

”A base at our University”

On 13 October 2017 Yu wrote to Epstein from San Francisco. The message is the clearest statement in the file of what he wanted the relationship to become:

Would like to continue the conversation I had with you regarding the developmental model I presented. I have an understanding of people like Josha and potentially other people that you support and would like to develop a base at our University where they could thrive, develop new stuff, and also be studied if they are open to it. See the attached, though I’d like to talk you through it.

“Josha” is Joscha Bach, who had been coming to Epstein’s house with Yu since 2015. The attachment is named on the sheet and is not in the release. The same message goes on to discuss rebranding Epstein’s island: “I have been talking to people close to David Copperfield. He also has two islands in the Bahamas… There is a group called ‘Unforbidden’ who organise events and are looking for an island to collaborate with… Could be a good rebrand for your island.” Epstein’s reply that morning was five words: “yes. or florida. palm beach.”¹⁰

Yu had described the model itself in November 2017, in the same message that accounted for the 100K. He wrote of giving talks to groups “involved in a practice I don’t condone, ‘psychedelic awakening’”, of being “able to ‘work’ on two of the subjects with very good results”, and that “They were ‘stage 3’ from the model that I presented to you via Skype”. He gave a link to the model, which is not in the release.³³ What the stages are, and what “work on” meant, is not stated on any sheet read for this page.

What Epstein said about him

Epstein’s private view of Yu is on the record twice, both times to people Yu had introduced or was about to meet.

On 5 November 2017, writing to the venture capitalist Masha Drokova ahead of an event of Yu’s: “gino says he will be with you, careful, he is a bit of a huckster”. Drokova replied a week later, across three paragraphs of one message: “I thought about what you said me about Gino as he talks to me a lot.” / “I think what makes it uncomfortable is that he wants to influence people and he is saying different things about people, directly to them and without them… he loves power” / “But I still hope that overall he has good intentions and wants to help people. on his own way” Between those two messages Epstein had written to her, “im alwsy very happy to see or help you , he is trying to get in under my good feelings for you.”³

On 4 January 2018, replying to the physician Deepak Chopra, who had offered to support a travelling event Yu was mounting in London, Epstein wrote nine words: “gino is a bit too mjuch of a hustler”.¹⁴

Both are Epstein’s characterisations of a man, written privately, and neither is a finding of this page. They are quoted because they are dated, they are his, and they bear directly on how he valued the traffic Yu was sending him.

Shanghai, and the conference that came apart

In August 2016 Yu told Epstein, “I am hosting TSC next year in Shanghai (http://www.consciousness.arizona.edu) and worked with a Mainland group to acquire the conference.”³¹ It did not happen. On 30 March 2017 Stuart Hameroff, whose Center for Consciousness Studies runs that conference, gave Epstein his own account at Epstein’s request. He described putting out “35 to 50k for speakers (Including 9k for Roger Penrose business class from London)”, a promised $100,000 a year for five years whose “First payment 100k was to be March 1. Didn’t happen”, a contract demanding exclusive intellectual-property rights for twenty years on a midnight deadline over a weekend, the mainland group writing to all his speakers under its own logo and “asking for credit card and bank account information in an open email”, and a threat that “our permit would be blocked by their government connections”. Yu appears twice:

The Woodenfish conference was something Gino tried to tag onto our conference which would have been OK except they advertised all our speakers without asking them, or us (who would have paid their expenses).

Gino denies any knowledge but he’s been involved the whole time in some way.

Hameroff moved the conference to San Diego and ended by asking Epstein, “Wanna help?” The next day Epstein answered: “who are the speakers and what is the cost. Im in for at least 50k, before knowing anyhing”.³⁵

The last two years

The traffic thins. Yu’s own sending falls from 130 dated documents in 2017 to 41 in 2018 and 3 in 2019, and other people begin managing what he had managed.

In November 2017 the near-death researcher Peter Fenwick, whose experiments Epstein was funding, wrote after a warning from Epstein that “unless you take control I must stop =ny serious consideration of continued support”: “Entirely agree, I have dropped Gino from the =eam and will make sure that I control the =thers.” Those = characters are on the face of the sheet, not in the extraction alone: the message reached the release through an export that turned soft line breaks into printed = signs, and no undamaged copy of it survives in this release. The words are “team” and “others” and “any”, and they are not supplied above.³⁶ The following August, refusing Fenwick a further twenty thousand dollars, Epstein returned to the same subject: “reminder when i asked gino for his breakdown was when i fo=nd out 50% went for drinks and food”, carrying the same damage, on the same kind of export, in the word “found”.³⁷

Yu was still supplying people to the end. On 16 and 17 July 2018 Groff was arranging a one-thirty meeting at the house between Epstein, Yu and a man named Mitch, down to whether to serve snacks.³⁸ In September 2018 he wrote from Zurich, “Here to sign the space in Davos.”³⁹ In January 2019 he shared a chalet in Klosters with Joscha Bach during the World Economic Forum, whose invoice Epstein’s office paid: Bella Klein wrote on 28 January, “ok, its paid already.”⁴⁰ The last of his messages in this file, dated 31 January 2019, introduces a man he had met at Davos and pastes his biography; the one before it, two days earlier, is headed “Introductions”.⁵

One message from that January is his own, image-verified, and should be read as it stands. On 11 January 2019, writing to Joscha Bach and copying Lesley Groff, Yu added to the Davos arrangements: “I will also have a shaman with me as well as a psychonaut that grows his own mushrooms.” Beneath it, Bach’s message of that morning to Groff reads: “Gino Yu tells me he would be most happy if ███ wants to join our accommodations in Davos, and also offers to take her along to Zurich and Marrakech after WEF. I am sure she would have a lot of fun. (Jeffrey already met with Gino).” The sheet was rendered locally at 300 dpi and read, and the producing party has covered one name in that sentence with a black bar. Nothing is supplied here in its place, and this page draws no inference about who it is.⁴¹

What the outside record says, and what it does not

The correspondence became public with the February 2026 release of the final tranche of Justice Department records, and was reported in Hong Kong. Dimsum Daily, drawing on a review by the outlet HK01, reported that Yu, a retired Polytechnic University professor, first met Epstein in 2015, exchanged at least 548 emails with him to 2019, and proposed developing a university “base”. The same report states: “Reporters have sought comment from Gino Yu but, as of publication, he had not responded.”¹⁵

That is where the outside record stops. No statement by him, and none by his university, was returned by the searches made for this page. A silence is not an answer and this page does not treat it as one, but it is recorded because on every comparable page the subject’s own account is set beside the documents, and here there is none to set.

What this page does not establish

The 209 forename documents are his by reading, not by query. Each was opened and the line carrying the string was read. A reader who re-runs "Gino" and subtracts every form of the full name will get 359 documents, not 209; the assignment above is a judgement about each one, and the table is printed so that it can be overruled.

The payment record here is an instruction record. The file shows Epstein offering $100,000, saying he had sent $25,000 to a conference, and instructing $25,000 out of his foundation. It contains no bank confirmation for any of them, and unlike the Kuhn payments there is no outside filing that closes the loop.

Nothing here establishes that the conduit Yu offered was ever used. The non-profit whose address he gave Epstein on 9 March 2018 exists and is easily identified: hinmancharity.org resolves to The Hinman Charity, which describes itself as established in 2013 out of a foundation of 2004, gives its purpose as “Preserving Mystical Cultures” and says it is guided by Buddhist principles. The release contains no trace of it beyond that one offer. The string hinmancharity returns exactly two documents in the whole index, and both are copies of Yu’s own message of 9 March 2018; "Hinman" returns 53 documents and not one of them is inside this population; "Hinman Charity" and "Hinman Foundation" return nothing at all. A non-profit named in an offer is not a non-profit that received anything, and no document read for this page shows that it was used, approved, or mentioned again by anyone. The charity is named here because the subject named it himself in a sentence this page quotes; the individual its website names is not named here, because he appears nowhere in the release and nothing is alleged against him.¹³⁴²

Nothing here establishes what the “developmental model” was. Yu’s own messages refer to stages, to a Skype presentation and to an attachment; none of those is in the release, and this page does not speculate about their content.

Nothing here establishes an ending. Twelve forms that would mark a rupture were searched against the whole index and intersected with this population. Eleven return nothing inside it: "return the funds", "return the money", "repay", "repaid", "returning the", "give back the", "not proceed", "wind down", "stop funding", "no longer involved", "terminate the". The twelfth, "refund", returns three, and they are printed here rather than left for a reader to find. Two are copies of the notice in which Stuart Hameroff’s Center for Consciousness Studies told its registrants that the Shanghai co-host had collected their money and that the Center would reimburse them; the third is a book-deposit refund on a study-course account. None is a rupture between Epstein and this subject, so the conclusion stands, but eleven empty intersections are the negative, not twelve. Nor does the population run on: of the 1,302 documents, 1,271 yield a year in the first fourteen non-blank lines of their first sheet, and not one of those years is later than 2019 (2015: 192, 2016: 462, 2017: 367, 2018: 187, 2019: 45, and 18 earlier). No document read for this page records what became of the relationship, and this page does not assert that one exists.⁶⁴³

Sources

Every quotation below is a single contiguous span from one sheet. Where a sheet was rendered locally from the released PDF and read at the image, the footnote says so and gives the resolution; where it was not, the footnote says the text layer only. Where a value is covered by a black bar, nothing is supplied in its place. Message addresses and telephone lines are not reproduced.

Four conventions, stated so that a reader testing a quotation against a sheet knows what to allow. Runs of spaces and a line break inside a quoted span are collapsed to one space, and a line break in an address block or a title page is written ” / ”. An elision inside a span is marked ”…” and nothing else is dropped. Where a sheet uses double quotation marks inside a passage this page is itself quoting, they are printed as single marks; that is nesting, never transcription, and no other punctuation is changed. Quoted-printable damage, a printed = standing where a letter should be, is left exactly as the sheet prints it and is never repaired, because on these exports the = is on the face of the page and not an artefact of the extraction.

Related

No link below asserts a relationship beyond what the cited sheet shows.

• The chain he began: Robert Lawrence Kuhn, whom he introduced to Epstein in August 2016 at Epstein’s request. That page carries the whole of what followed, including the payments from Gratitude America, Ltd., and this page carries only the introduction.
• The money: Gratitude America, Ltd., the foundation named in the instruction “send gino 25k from gratitiude”; Richard Kahn, the accountant the instruction was addressed to; HBRK Associates Inc., the firm in Kahn’s signature block on his reply to that instruction.
• People named on sheets cited here: Jeffrey Epstein (the correspondent throughout); Lesley Groff (who booked every meeting quoted above and took delivery of the headset arrangements); Bella Klein (who confirmed the Davos accommodation was paid); Karyna Shuliak (named in the message of 7 August 2016); Joi Ito and Leon Black (each named on one of the two schedules rendered and read for this page, beside the entries that carry him).
• Places: 9 East 71st Street, the New York house where the headset was installed and where Epstein’s office booked him in; Little Saint James, which is the island the message of 13 October 2017 calls “your island”; the sheet gives no name for it and none is supplied here.
• Topics: Calendars and daily schedules, whose own text notes that appointment records establish only that meetings were scheduled or planned; Epstein office monthly schedules, the series from which the November 2015 and October 2016 sheets come; Epstein and academic institutions and Epstein network: academia, for the wider pattern of science funding of which this is one instance.

Footnotes

1. Letter on Novamente LLC letterhead (“intelligent virtual agents”), signed “Ben Goertzel, / Novamente LLC” over a handwritten signature, addressed to “Dr. Gino Yu / Associate Professor / Multimedia Innovation Centre, School of Design / Hong Kong Polytechnic University / Hung Hom, Hong Kong”, supporting the proposal “A Software Toolkit for Creating Intelligent Non-Player Characters in Video Games” with “HK$280,000 in cash as matching fund sponsorship” and referring to an office in Xiamen “that we intend to open in late 2010”. The letter carries no date. Read at the image at 500 dpi by this pass on 2026-09-12 (and at 300 dpi by the writer). Two lines of the address block below the city are covered by black bars and nothing is supplied here in their place. The office the letter says Novamente intends to open in late 2010 is in Xiamen, China, not in Hong Kong; an earlier draft of this page called it a Hong Kong office in the infobox. The letter also says the firm will apply the system “in our US and Brazilian offices”. https://epstein-data.com/EFTA00728504 p.1. ↩ ↩² ↩³ ↩⁴

2. “OpenCog / AGI Toddler Project / Open-Source ‘Human Toddler Level’ AGI by 2015 for US $3M Total / Ben Goertzel and Gino Yu, Principal Investigators / Multimedia Innovation Centre, P507 Hong Kong Polytechnic University / Hung Hom, Hong Kong”, marked “Rough draft, not yet for distribution”. The management-board sentence stands on printed p.16. Text layer only. https://epstein-data.com/EFTA01114145 p.1; https://epstein-data.com/EFTA01114145 p.16. The run is 19 sheets. ↩ ↩² ↩³

3. Masha Drokova to Epstein, “Subject: Re:”, “Date: Sun, 12 Nov 2017 16:47:03 +0000”, carrying Epstein’s messages of “Nov 5, 2017, at 2:54 AM” (“gino says he will be with you, careful, he is a bit of a huckster”) and “Nov 5, 2017, at 6:23 PM”, and her own of 5 November 1:17 PM. Read at the image at 500 dpi by this pass on 2026-09-12. Her address is barred at two points. The sheet prints “he loves power” with no full stop, at the end of a paragraph, and “on his own way” with no full stop, at the end of the next; an earlier draft of this page supplied both stops and ran the three paragraphs together as one span. Epstein’s “gino says he will be with you, careful, he is a bit of a huckster” also prints with no terminal stop, and is quoted here without one. https://epstein-data.com/EFTA00956461 p.1. Epstein’s message of 5 November standing alone, addressed to her, is at https://epstein-data.com/EFTA02570560 p.1 and https://epstein-data.com/EFTA00962809 p.1. ↩ ↩² ↩³

4. “Creating Robots with Toddler-Level Intelligence Using the OpenCog AGI Architecture / Research Proposal for the Epstein Foundation / Ben Goertzel / February, 2013”. Text layer only. The title block stands on printed p.1 and the ITF sentence naming Yu as principal investigator on printed p.2, and both sheets were read: https://epstein-data.com/EFTA01103509 p.1; https://epstein-data.com/EFTA01103509 p.2. The run is 10 sheets. Because this footnote names two sheets of the same run, a mechanical test of its pins cannot fail them; the split is stated here so a reader can check it by eye. The companion document headed “Jeffrey Epstein IV Foundation / AGI Initiative / R&D Funding Proposal / Ben Goertzel / December 2, 2013” is at https://epstein-data.com/EFTA01103465 p.1, a 30-sheet run. It names this subject once, in its bibliography and nowhere in its body, which is why it falls inside the 1,086-document base phrase. Printed p.30 carries reference [13]: “Ben Goertzel, Joel Pitt, Zhenhua Cai, Jared Wigmore, Deheng Huang, Nil Geisweiller, Ruiting Lian, and Gino Yu. Integrative general intelligence for controlling game ai in a minecraft-like environment. In Proc. of BICA 2011, 2011.” That sheet was read at the image at 400 dpi by this pass on 2026-09-12: https://epstein-data.com/EFTA01103465 p.30. The title block at p.1 is text layer only. An earlier draft of this footnote said the document “does not name this subject”, which was false and is corrected here rather than quietly removed. ↩ ↩² ↩³

5. Gino Yu, “Sent: Thursday, January 31, 2019 9:45 AM”, to [email protected], subject “Another one who is interesting to meet.”, body “Met in Davos. Let me know if you’d like to meet him.” followed by a pasted biography. Read at the image at 400 dpi by this pass on 2026-09-12: the biography is riddled with quoted-printable damage on the face (“think t=nk”, “humanit=rian”, “Davos/=losters”) and the two quoted spans print undamaged. Yu’s address is barred; the recipient value is Epstein’s own mailbox, unbarred. The man named in the biography is a living private individual against whom nothing is alleged on any sheet read here, and he is not named on this page. https://epstein-data.com/EFTA02628503 p.1. His message of two days earlier, “Sent: Tuesday, January 29, 2019 11:50 AM”, subject “Introductions”, is at https://epstein-data.com/EFTA02626861 p.1. ↩ ↩² ↩³

6. Every count on this page is of distinct documents, not pages, taken on 2026-09-12 with exact-phrase FTS5 queries against the corpus full-text index, which covers the EFTA, HOUSE_OVERSIGHT and DOJ-OGR productions together. LIKE was not used anywhere. The forms tried and what each returned: "Gino Yu" 1,086 · "Gino Y" 20 (5 outside the base) · "Dr Gino Yu" and "Dr. Gino Yu" 19 (0) · "Yu Gino" 11 (0) · "Guno Yu" 6 (0) · "ginoyu" 2 (0) · "Gino Vu" 1 (1) · "Gino Yn" 1 (1). Union of all forms: 1,093. Forms tried that returned nothing: "Gin0 Yu", "Gina Yu", "Gmo Yu", "Gino Yo", "Ginu Yu", "Glno Yu", "G1no Yu", "Cino Yu", "Gino M Yu", "Ginp Yu", "Gino Ya", "Gino Yui", "Gino Yuu", "Gino Xu", "Gino Wu", "Gino Hu", "Gino Lu", "Gino Yu PhD". Surrounding counts: bare "Gino" 1,452 · bare "Yu" 2,194 · "benefactor" 259 corpus-wide and 39 inside this population · "Woodenfish" 17, all inside · "OpenCog" 318 corpus-wide and 26 inside · "Lydia Laurenson" 44 and 17 inside · "consciousness hacking" 35 and 15 inside. scripts/ocr_variants.py was run on "Gino Yu" and produced one candidate, "Gin0 Yu", which returns zero documents; every damaged form that exists in this release was found by reading. ↩ ↩² ↩³ ↩⁴ ↩⁵

7. The decomposition of the 359. Bare "Gino" returns 1,452 documents and the all-forms union accounts for 1,093, leaving 359. All 359 were opened and the line carrying the string was read in each; the bucket totals in the table above are 209 + 54 + 28 + 14 + 12 + 12 + 10 + 6 + 4 + 4 + 2 + 2 + 1 + 1 = 359. A reconciling total proves the buckets partition; it does not prove that any one of them contains what its name says, so each assignment is a judgement about a document that was read, and the buckets are listed so a reader can overrule them. The 28 Severini documents are art-collection schedules and auction records listing “SEVERINI, GINO / L’autobus” and “Il Tango Argentino”. The 14 Cicerchia documents are social-network notifications and two direct messages from a Palm Beach estate agent, 2010 to 2013. The 12 Segrè documents are mailing-list address blocks in which the physicist’s name sits between other recipients. The 12 “other people and businesses” carry the architect Gino Valle, the lyricist Gino Paoli, the music critic Gino Castaldo, the Riva partner Gino Gervasoni, a contact card reading “gino ong”, a donor ledger line “Gino, George”, an enforcement report naming Gino F. Scibelli, a payroll circulation list naming Gino Alcoba, a hotel charge in which the Fairmont Grand Del Mar is rendered “FAIRMONT GINO DELMAR”, and two garbled magazine credits. The 10 Vizzari documents are copies of one jail housing and population list. The 6 Silvestri documents are Western Union subpoena returns addressed to “GINO SILVESTRI / MONEY LAUNDERING TASK FORCE / STATE ATTORNEYS OFFICE”. The 4 “Gino Avenue” documents are a law firm’s Coconut Grove street address on pleading footers. The 4 Francesca Gino documents quote a management-studies paper. The 54 noise documents carry no name at all: bank and telephone statements, freight-label tables, equipment manuals, and, four times, a form’s tick-box pair ”☐ Yes ☑ No” which the extraction renders as a single token — giNo on https://epstein-data.com/EFTA00166918 p.8, https://epstein-data.com/EFTA00167035 p.4 and https://epstein-data.com/EFTA00168803 p.9, and GiNo, with a capital G, on https://epstein-data.com/EFTA01284060 p.10, all four of them “DB PWM GLOBAL KYC/NCA” account-opening forms. The box was read at the image: https://epstein-data.com/EFTA00166918 p.8 was rendered at 500 dpi for this pass on 2026-09-12 and prints, against each of four questions, an empty square labelled “Yes” beside a ticked square labelled “No” — the pair the extraction collapses into one token. Nothing else on that sheet is described here; it is an account form carrying personal identifiers, several of them barred by the producing party, and no value from it is quoted or characterised. A fifth document carrying the token giNo, https://epstein-data.com/EFTA00390915 p.5, is not a tick-box pair at all: it is a retail advertisement whose line reads “giNo Sales Tax! - Most Orders are Tax Free”. An earlier draft of this page counted it as a fifth tick-box. It stays in the 54 either way, because it carries no name, so the bucket totals above are unchanged. All five text layer only. ↩ ↩² ↩³

8. The top-level sender measurement, run on 2026-09-12 over the first sheet of each of the 1,302 documents, under the rule stated in the body. Counters at a twelve-non-blank-line window: sender parsed 1,069, sender value absent, barred or blank 187, no sender label 46; his name in the sender value 330, with his forename at the head of the value in 330 of 330; addressed to Epstein in 291 of the 330; addressed without authoring 332. At windows of 8, 20 and 30 non-blank lines the sender figure is 330, 330 and 330. Over the 1,086-document base phrase alone the same rule returns 326, also stable at all four windows. scripts/act_rate.py --selftest printed SELFTEST PASS before any figure from it was taken; act_rate.py '"Gino Yu"' "Gino Yu" "Gino" returns sender-field act 405 of 1,086 (37%), addressed 321, header-less authored 8, press 0, PARSER REFUSED 26. A PARSER REFUSED line is a refusal, not a negative, and none of those documents is counted against him anywhere on this page. Of the 85 documents the tool scores and this rule does not, 50 are sheets on which the tool has joined two or more sender labels with |, and on 67 of the 85 the first sender label on the sheet belongs to somebody else. The recipient figures, under the recipient rule stated in the body. Addressed to Epstein: 287 of the 326 on the base phrase, 290 of the 329 on the all-forms union, 291 of the 330 over the whole population. Addressed without authoring: 326 on the base phrase, 326 on the union and 332 over the whole population. That last figure moves by one on a single judgement about the 209: https://epstein-data.com/EFTA02105141 p.1, a message of May 2014 whose To: value the extraction renders “Gino oils’”, is read here as somebody else, because it predates every other trace of this subject in the release by a year and nothing on the sheet identifies him; a reader who reads it as him gets 333. An earlier draft of this page printed 323 and 329 here, from a word-boundary test alone. The three documents a word-boundary test cannot reach are https://epstein-data.com/EFTA00325639 p.1 (To:GinoYu<ta), https://epstein-data.com/EFTA00437949 p.1 (To:GinoYu<ta) and https://epstein-data.com/EFTA02191765 p.1 (GinoYul), each a Lesley Groff message whose body addresses him by name (“Hi Gino…”, “Hello Gino…Jeffrey is asking if you could come see him”, “On Oct 25, 2016, at 11:49 AM, Gino Yu”); all three are read in the text layer only, the rendering quoted is the extraction’s and not a claim about what the sheets print, and all three lie inside the 1,086-document base phrase. The base and the union return the same 326 because none of the seven documents the damaged spellings add is addressed without authoring. The three the mailbox test reaches only by ignoring the domain are https://epstein-data.com/EFTA00958429 p.1, https://epstein-data.com/EFTA01739083 p.1 and https://epstein-data.com/EFTA01740106 p.1, also text layer only. Both tests are now printed in the body, because a figure whose rule is published and whose name test is not is still a figure a reader cannot reproduce: three implementations of this rule returned 323, 323 and 326 on the union, and the whole of that difference was the name test, not the rule. Duplicate analysis over the 330, under the timestamp rule stated in the body: 315 yield a parsable top-level timestamp and 15 do not; those 315 carry 290 distinct timestamps with offsets converted to UTC (24 timestamps shared by 49 documents) and 307 compared as printed (8 shared by 16). Year tally over the 315: 2015 16, 2016 125, 2017 130, 2018 41, 2019 3. An earlier draft of this page printed 316 / 297 / 14 and a range of 297 to 311 with no timestamp rule published, and printed 329 and 291 with no recipient rule published — which is the same fault this page objects to in a number quoted without its rule. A second draft published the recipient rule but not its name tests, and printed 323 and 329. The rule and both name tests are now published, and every figure here is what they return. ↩ ↩² ↩³ ↩⁴

9. Epstein to Gino Yu, subject “Re: Do you have a Vive?”, “Date: Sat, 22 Oct 2016 01:36:05 +0000”, body “Donot have”, carrying Yu’s message of “Friday, 21 October 2016”. Read at the image at 300 dpi by the writer on 2026-09-12. Yu’s address is covered by a black bar at both points it appears. https://epstein-data.com/EFTA00814303 p.1. Lesley Groff to Epstein, subject “where for Gino to set up tomorrow?”, “Sent Tuesday, October 25, 2016 3:30:45 PM”, carrying Yu’s message beneath it, https://epstein-data.com/EFTA01739080 p.1; Epstein’s answer “small room off library”, “Sent Tue 10/25/2016 3:32:18 PM”, https://epstein-data.com/EFTA02191668 p.1; the schedule entry “(TBD Gino Yu to arrive and set up VIVE in the small room off of the library)”, https://epstein-data.com/EFTA01739251 p.1; Groff to Epstein and Merwin Dela Cruz, “Sent: Wednesday, October 26, 2016 7:27 PM”, subject “Gino is in LA! 4pm is cancelled!”, https://epstein-data.com/EFTA02447398 p.1. That last sheet was read at the image at 500 dpi by this pass on 2026-09-12: it prints “for next =eek!!” — a quoted-printable soft break on the face of the sheet — and an earlier draft of this page printed “week” and ran the subject line and the body together as one quotation. Groff’s address is barred. The other three text layer only. ↩ ↩² ↩³ ↩⁴ ↩⁵

10. Gino Yu to Epstein, subject “Re: Cuba Update”, “Date: Fri, 13 Oct 2017 12:34:10 +0000”, his own reply of three minutes later, which carries Epstein’s “yes. or florida. palm beach.” of 12:31:45 and, beneath that, his own message of “Fri, Oct 13, 2017 at 2:07 PM” in full and undamaged. Read at the image at 500 dpi by this pass on 2026-09-12. His address is covered by a black bar at three points. https://epstein-data.com/EFTA00991360 p.1. Every quotation in this section is a single contiguous span of that one sheet, and the sheet was chosen for that reason. Three copies of the passage survive and the other two both damage it: on Yu’s own sheet, https://epstein-data.com/EFTA00991316 p.1, the extraction runs two words together as “Universituvhere”; on Epstein’s reply, https://epstein-data.com/EFTA00991353 p.1, the words “I’d” are missing altogether, the extraction reading “See the attached, though • / like to talk you through it.” An earlier draft of this page quoted the full sentence while pinning it to that second copy, which would have been a span of two sheets presented as one; the pin is now on the sheet that carries the whole of it. Both copies text layer only. ↩ ↩²

11. Gino Yu to Epstein, “For the Cuba conference, they are looking for the sponsorship (25K). Who should I direct them to? This is their current writeup (attached), but I intend to edit.”, carrying Epstein’s message of “30 Jan 2017, at 8:00 AM” beneath it, https://epstein-data.com/EFTA01054913 p.1, with a second copy at https://epstein-data.com/EFTA02661895 p.1; and “The Cuba group are asking about the funding. Is it possible to transfer 25K for the event?”, https://epstein-data.com/EFTA01053550 p.1, with copies at https://epstein-data.com/EFTA01054265 p.1, https://epstein-data.com/EFTA02377372 p.1 and https://epstein-data.com/EFTA02660077 p.1. All text layer only. The conference programme, “International Conference / FROM KNOW TO NEW: CREATIVITY AND INNOVATION IN NEUROSCIENCE / La Habana (Cuba), 6 - 8 Novembre 2017”, is at https://epstein-data.com/EFTA00587664 p.1. ↩ ↩²

12. Epstein to Gino Yu, “Sent: Wednesday, March 15, 2017 4:51 PM”. Three copies of that line survive and two of them damage it. Epstein’s own sheet, https://epstein-data.com/EFTA02656417 p.1, read at the image at 400 dpi by this pass on 2026-09-12, prints “i have sent 25k alr=ady” — a quoted-printable soft break on the face of the sheet, with the same damage running through its confidentiality footer (“commu=ication”, “addre=see”). A second copy, https://epstein-data.com/EFTA02656612 p.1, prints “i need a=I details on cuba”. The line is therefore quoted above from the third copy, Yu’s reply of 18:19:34 the same day, https://epstein-data.com/EFTA01052096 p.1, which prints it in full and undamaged: “i need all details on cuba i have sent 25k already”. Read at the image at 400 dpi by this pass. Yu’s address is barred there. An earlier draft of this page pinned the quotation to EFTA02656417 and said in this footnote that the quoted span printed undamaged there. Neither the image nor the text layer of that sheet carries the word “already”: the undamaged reading only ever existed on the copy the earlier draft cited last and described, wrongly, as “Yu’s earlier message” — it is his reply. The second copy is text layer only. ↩ ↩²

13. Epstein to Richard Kahn and Gino Yu, no subject, “Date: Thu, 08 Mar 2018 21:06:11 +0000”, body “rich , send gino 25k from gratitiude. forundation. so he can find specali ability people”. Read at the image at 300 dpi by the writer on 2026-09-12. Both recipients’ addresses are covered by black bars. https://epstein-data.com/EFTA00862301 p.1. Further copies of the same instruction, each quoted inside Kahn’s or Yu’s reply, are at https://epstein-data.com/EFTA00862378 p.1, https://epstein-data.com/EFTA00862383 p.1, https://epstein-data.com/EFTA02508157 p.1 and https://epstein-data.com/EFTA02509436 p.1. Yu’s reply, copying Kahn, “Subject: Re:”, “Date: Fri, 09 Mar 2018 18:45:43 +0000”, offering the Hong Kong account and the United States non-profit, is at https://epstein-data.com/EFTA00863609 p.1. Read at the image at 600 dpi by this pass on 2026-09-12. The sheet prints the address as (http://hinmancharity.org/about/), a live hyperlink with a trailing slash, followed by three spaces and “Would the non-profit be of benefit to you for tax purposes?”. The text layer of that sheet renders the same span (http://hinmancharity.orgiaboutO Would the non-profit, collapsing the slash, the closing bracket and the space; an earlier draft of this page reconstructed the address from that text layer, supplied a slash and a bracket, and dropped the trailing slash the sheet prints. The address above is now transcribed from the image. Yu’s own address and Kahn’s are covered by black bars on this sheet; Epstein’s own mailbox is not barred, and it is the address this archive’s subject used throughout. A second copy of the same reply, https://epstein-data.com/EFTA02509724 p.1, carries quoted-printable damage throughout and renders the address (http://hinmancharity.orgiaboutt); it is text layer only and nothing is quoted from it. Epstein’s question to Robert Kuhn of 15 February 2017, “is it a charity that received the funding?”, is cited on his page and is not re-cited here. ↩ ↩² ↩³

14. Epstein to Deepak Chopra, “Sent: Thursday, January 4, 2018 2:26 PM”, subject “Re: Evolving Caravan”, body “gino is a bit too mjuch of a hustler”, carrying Chopra’s message of that morning and Yu’s of “Thursday, January 4, 2018 7:09:06 AM”. Read at the image at 400 dpi by this pass on 2026-09-12. The sheet carries the same quoted-printable damage as the other exports in this set (“c=n support”, “=ino Yu”, “li=e to start”) and the nine quoted words print undamaged — which this pass established by opening the sheet, not by inference from the text layer. Chopra’s contact block is covered by three black bars. https://epstein-data.com/EFTA02545218 p.1. ↩ ↩²

15. “Retired PolyU Professor Gino Yu linked to frequent contact with Jeffrey Epstein in new U.S. files”, Dimsum Daily, February 2026, https://www.dimsumdaily.hk/retired-polyu-professor-gino-yu-linked-to-frequent-contact-with-jeffrey-epstein-in-new-u-s-files/, which attributes its review of the records to the Hong Kong outlet HK01 and reports “at least 548 emails” exchanged between 2015 and 2019 and that “Reporters have sought comment from Gino Yu but, as of publication, he had not responded.” A companion piece, “What exactly was retired PolyU Professor Gino Yu trying to develop with Epstein?”, https://www.dimsumdaily.hk/what-exactly-was-retired-polyu-professor-gino-yu-trying-to-develop-with-epstein/, describes the university “base” proposal in the same terms as the document at ¹⁰ and cites no named source of its own. These are outside sources, not documents in the release, and they are cited as the report of a review rather than as primary records. No statement by this subject and no statement by his university was returned by the searches made for this page. ↩ ↩²

16. The seven documents added by the damaged spellings, over and above the 1,086-document base, were each opened and each is this subject: https://epstein-data.com/EFTA02399449 (sender field “Gino Vu”, “Greetings from Agra.”), https://epstein-data.com/EFTA01739237 (an office schedule printing “4:00pm Appt w/Gino Yu”, the extraction reading “Gino Yn”), https://epstein-data.com/EFTA00285556 (a schedule line the extraction reads as “5:30pm Appt w/Gino Y”), https://epstein-data.com/EFTA02040595, https://epstein-data.com/EFTA02329562, https://epstein-data.com/EFTA02567641 and https://epstein-data.com/EFTA02650552 (on each of which the extraction reads “Gino Y” or “Gino y.” where a sender value stands), each p.1. One of the seven, EFTA01739237, was read at the image and is set out below; the other six were read in the text layer only, and what those six sheets print was not checked at the image and no claim is made about it here. (An earlier draft of this footnote said “those five sheets”, which counted neither the six nor the four named in the parenthetical above.) EFTA01739237 was read at the image at 500 dpi by this pass on 2026-09-12 (and at 300 dpi by the writer): the sheet is a Lesley Groff schedule whose header reads “Sent: Tuesday, October 25, 2016 1:38:18 PM”, and it prints “4:00pm Appt w/Gino Yu” followed by a black bar where a contact value would stand. The text layer of that sheet reads “Gino Yn” and shows no bar at all. It also drops the Sent: value, and it drops the whole Cc: line, on which the sheet prints two further bars. That is the whole case for going to the image. ↩

17. The bare-surname test. "Yu" returns 2,194 documents, of which 1,093 are in the all-forms union above and 3 more are among the forename-only documents, leaving 1,105. Every one of the 1,105 was scanned at the line carrying the token and bucketed by the word before it. The largest identifiable person is Philip Yu, whose forename stands immediately before the token in 62 of them, with a further 11 reading “Cortese-Philip Yu” or “Cortese/Philip Yu”, for example https://epstein-data.com/EFTA00304799 p.2 (“Call Philip Yu between 2-3:00”) and https://epstein-data.com/EFTA01843001 p.1 (subject “Re: Rich cortese-Philip Yu”). Others carrying a preceding word are Yu Myung-hwan (19), New York rendered “New Yu*” (17), David Yu (10), Ki Jun Yu (8), Mindy Yu (7), Glenda Yu (7), Vagit Yu Alekperov (6), Julie Yu (4), Soon Yu (4) and Lin Yu (2). 494 of them carry no capitalised Yu token at all, and the dominant cause is the extraction rendering the word you as yu: https://epstein-data.com/EFTA00714986 p.1 (“do yu have”), https://epstein-data.com/EFTA00993078 p.1 (“where are yu”), https://epstein-data.com/EFTA01017384 p.1 (“see yu tomorrow”), https://epstein-data.com/EFTA00629407 p.1 (subject “Re: Where can I cal yu”), https://epstein-data.com/EFTA00373529 p.1 (“please confirm Jeffrey is to call yu on the number”), each read in the text layer. No figure on this page is keyed on the surname. ↩

18. Epstein’s own forename shorthand, each read in the text layer and each outside every form of the full name: “Ask Gino if we can also do another five on wed”, 28 March 2016, https://epstein-data.com/EFTA00327392 p.1 and https://epstein-data.com/EFTA02055564 p.1 · “why dont you ask gino , i think one of the conferences he runs is a charity in the states. he can transmit the money to you”, to Peter Fenwick and Alain Forget, 26 December 2016, https://epstein-data.com/EFTA01058077 p.1, and Forget’s reply “peter runs the show; gino why not?”, https://epstein-data.com/EFTA01058082 p.1, and Fenwick’s message of the next day under the subject “Gino and grant”, https://epstein-data.com/EFTA01058189 p.1 · “remind me james tv/ / sunday gino and deepak.”, 10 November 2017, https://epstein-data.com/EFTA00462744 p.1 · “Gino Miro Wolfe Leon”, 4 March 2018, https://epstein-data.com/EFTA00469710 p.1 · “ehud , at 1 , michael wolf and terje to join at 2 . . gino? this morning? leon? miro ? maybe tonite?”, 5 March 2018, https://epstein-data.com/EFTA00469681 p.1 · “thnx gino. we met at leon black with milken. email when you return to new york .”, 17 May 2018, https://epstein-data.com/EFTA01058540 p.1. Two of Epstein’s own calendar alert exports carry the entry text “gino new york” for 5 July 2016, https://epstein-data.com/EFTA02456961 p.1 and https://epstein-data.com/EFTA02458034 p.1. ↩

19. Epstein office schedule for Wednesday 18 to Friday 20 November 2015, “Date: Tue, 17 Nov 2015 17:51:08 +0000”. Read at the image at 500 dpi by this pass on 2026-09-12 (and at 300 dpi by the writer): the sheet prints, in italic, “Reminder: George Mueller (Gino Yu’s friend) to arrive NY,” followed by a black bar, while the text layer of that same sheet reads “Gino Nes friend” and shows no bar. The two entries that follow it print “8:00am Appt w/Justin Nelson” and “8:30/9:00am BREAKFAST w/Joi Ito”, each also followed by a bar, and “12:30pm LUNCH w/Misha Gromov” follows those. The sender value and the recipient value are barred, so the sheet does not itself name Lesley Groff; the text layer’s date line reads “The, 17 Nov 2015” and the image prints “Tue,”. Nothing barred is supplied here. https://epstein-data.com/EFTA00334985 p.1. Forty-eight documents carry the same reminder; 25 are reachable by some form of his name and 23 are not. The census normalisation, published so the rendering count can be reproduced: the sheet text is whitespace-collapsed, the three words are taken as everything between “Mueller” and “friend”, and leading punctuation is kept. On that rule the release carries 22 distinct renderings, of which 13 contain no Yu at all; treating , Gino Yu's as the same rendering as Gino Yu's gives 21, and an earlier draft of this page gave 23 and counted Gino Ms twice because one of its two documents breaks the string across a line. The thirteen surname-free renderings, and the 23 documents no name form reaches, are stable under all three normalisations. Other spellings read for this page, each in the text layer: https://epstein-data.com/EFTA00334421 p.1 (“Gino Pr’s”), https://epstein-data.com/EFTA00334427 p.1 (“Gino Ms”), https://epstein-data.com/EFTA00334484 p.1 (“Gino Pt’s”), https://epstein-data.com/EFTA00334956 p.1 (“Gino lit :s”), https://epstein-data.com/EFTA00334964 p.1 (“Gino Yith”), https://epstein-data.com/EFTA00334966 p.1 (“Gino Yies”), https://epstein-data.com/EFTA00335016 p.1 (“Gino Hes”), https://epstein-data.com/EFTA00335038 p.2 (“Gino Yir:s”), https://epstein-data.com/EFTA00335102 p.1 (“Gino Yit:s”), https://epstein-data.com/EFTA00335434 p.1 (“Gino Pis”) and https://epstein-data.com/EFTA00840597 p.1 (“Gino Yit:s”). None of those eleven is reachable by any form of his name. The corrected copy in the House Oversight production is at https://epstein-data.com/HOUSE_OVERSIGHT_002011 p.1; that store returns HTTP 403 for HOUSE_OVERSIGHT_* identifiers and it could not be rendered. ↩ ↩²

20. Message to Epstein, “Sent: Mon 3/28/2016 8:31:49 PM”, subject “Re: Gini in Woody’s office re Brad Karp screening Wed-Not possible”, whose body opens “Gino’s reply re “when next”: / It will be very unlikely that it will be before Cannes in May.” and which carries beneath it a message signed ”> Gini / > Ginevra Tamberi”. Read at the image at 500 dpi by this pass on 2026-09-12 (and at 300 dpi by the writer): the sheet itself prints “Gino’s”, with an unambiguous o, so this is the document’s own error and not the extraction’s. The text layer of the same sheet renders the Sent value “Mon 3/28/2018”; the image prints “Mon 3/28/2016”, and 28 March was a Monday in 2016 and a Wednesday in 2018. The sender value and two further values are covered by black bars. https://epstein-data.com/EFTA02057731 p.1. The barred sender is Lesley Groff, and the copy set establishes it without supplying anything from the bar. Both copies were read at the image at 400 dpi by this pass on 2026-09-12, and they carry her name in two different places, which an earlier draft of this footnote and of the body above described wrongly as one: the message she is answering, https://epstein-data.com/EFTA02057653 p.1, prints “From: jeffrey E.” — its sender value is Epstein’s, not hers — with “Sent Mon 3/28/2016 8:14:58 PM” and the body “When next ?”, and carries her name unbarred twice over: in its “To: Lesley Groff[” line, where the bar falls on the address alone, and in the quoted attribution line “On Monday, 28 March 2016, Lesley Groff <”, a bar, ”> wrote:”. Her own original, https://epstein-data.com/EFTA02466737 p.1, prints “From: Lesley Groff” followed by a bar over the address, “Sent: Monday, March 28, 2016 7:45 PM” and “To: Jefffrey Epstein” — the sheet prints the forename with three fs — and that is the one sheet of the three that carries her as its sender value. Both carry the same ”> Gini / > Ginevra Tamberi” signature and neither contains the string “Gino” anywhere. The correction does not weaken the identification: the sheet that does not name her as sender is addressed to her and names her, unbarred, as the author of the message being forwarded, which is a second and independent place for the name to stand. An earlier draft of this footnote also declared both sheets text layer only; both have now been opened. bates_twins.py finds no Bates twin for EFTA02057731, which is not proof that none exists — a production stamp the extraction dropped is invisible to it. The corroboration here comes from the thread, not from a twin, and it corroborates the reading without being able to correct the string the sheet prints. ↩

21. The press test, established by reading rather than by the tool. act_rate.py classes a document as press only if its expression fires inside the first 1,200 characters of the first sheet. It reports press 0 here. All 46 documents whose first sheet carries no sender label at all were opened, that being where an authored paper or a clipping would sit: 20 are Epstein office monthly and weekly schedules; 10 are papers, books and funding proposals by Ben Goertzel, including the OpenCog volumes and the proposals at ⁴; 7 are device-forensics exports of calendar, contact, call and chat tables, https://epstein-data.com/EFTA00506472 p.1, https://epstein-data.com/EFTA00506602 p.1, https://epstein-data.com/EFTA01065289 p.1, https://epstein-data.com/EFTA01207995 p.1, https://epstein-data.com/EFTA01209970 p.1, https://epstein-data.com/EFTA01213181 p.1 and https://epstein-data.com/EFTA01217941 p.1; 3 are copies of the roboticist David Hanson’s one-page biography; 2 are conference programmes, the Havana meeting at https://epstein-data.com/EFTA00587664 p.1 and a Shanghai meeting headed “Consciousness, Science, Technology, and Society / Shanghai, June 7-9, 2017” at https://epstein-data.com/EFTA00810775 p.1; 2 are Skype conversation logs, https://epstein-data.com/EFTA01612465 p.1 and https://epstein-data.com/EFTA01623392 p.1, the second of which is a log of two unanswered messages from him, “There” and “There?”; one is the sponsorship letter at ¹; and one is an office message whose sender field is barred, https://epstein-data.com/EFTA02191971 p.1. None is a press capture. 20 + 10 + 7 + 3 + 2 + 2 + 1 + 1 = 46. An earlier draft of this page gave the total as 43 in this footnote and 46 in ⁸ for the same quantity, and itemised 19 schedules, 4 forensics exports and 2 barred-sender messages. The figure under the rule at its own twelve-line window is 46, itemised as above. The documents that itemisation did not reach are the March 2018 office schedule EFTA00285556 and the forensics exports EFTA01209970, EFTA01213181 and EFTA01217941; all four were opened for this pass and none of them is press. Re-running the tool’s own expression over the whole first sheet rather than its 1,200-character window fires on 203 of the 1,302: on 202 the only marker matched is “All Rights Reserved”, which is the closing line of Epstein’s own confidentiality footer, and on the last it is the word “SECTION:” inside a technical book. Widening the window is not a better measure either, and the figures above are counts of documents read. ↩

22. Lesley Groff to Joscha Bach, “Sent Thur 5/7/2015 12:04:23 PM”, subject “Jeffrey Epstein”, body “Good morning Joscha. Might you and Gino be available to come see Jeffrey today? Perhaps 1:30? Please let me know. Thanks, Lesley”, https://epstein-data.com/EFTA02080022 p.1; Bach’s reply, “Sent Thur 5/7/2015 12:54:53 PM”, “Sure! I will ask Gino, too. Bests, Joscha”, https://epstein-data.com/EFTA02079030 p.1; Groff to Epstein, “Sent: Thursday, May 7, 2015 11:49 AM”, subject “Reminder Gino Yu and Joscha”, https://epstein-data.com/EFTA02502684 p.1. The first two are text layer only. The reminder is quoted above from Epstein’s reply of two minutes later, https://epstein-data.com/EFTA02079969 p.1 (“Sent: Thur 5/7/2015 11:51:46 AM”, body “yes”), which carries the sentence in full and undamaged. Read at the image at 400 dpi by this pass on 2026-09-12. It is quoted from that copy because Groff’s own sheet does not print it undamaged: EFTA02502684, also read at the image at 500 dpi for this pass, prints “before y=u leave” — a quoted-printable soft break carried onto the face of the sheet, not an extraction fault — and an earlier draft of this page repaired it silently to “you” while declaring that sheet text layer only. Neither sender value is barred, and an earlier draft of this footnote said both were. Both sheets were re-read at the image at 400 dpi by this pass on 2026-09-12: EFTA02079969 prints “From: jeffrey E.”, unbarred, and what is barred on that sheet is its To: value and the name in the quoted attribution line (“On Thu, May 7, 2015 at 7:48 AM,”, a bar, “wrote:”); EFTA02502684 prints “From: Lesley Groff” with her name in the clear and the bar falling on the address alone, above an unbarred “To: Jeffrey Epstein”. Nothing is supplied from either bar and no finding moves; the statement was simply false about both sheets, in opposite directions. ↩

23. Gino Yu, “Subject: Available tomorrow!!”, “Date: Thu, 07 May 2015 17:00:36 +0000”; the body is his name and “Sent from my iPad”. The recipient value is barred. Text layer only. https://epstein-data.com/EFTA00348079 p.1. ↩

24. Gino Yu, “Subject: Re: Please Forward to Jeffrey”, “Sent: Sun 5/10/2015 9:57:26 PM”, carrying his own message of “On May 10, 2015, at 8:48 AM” addressed to Epstein’s office and asking that it be forwarded. Read at the image at 500 dpi by this pass on 2026-09-12: the header labels print with colons where the text layer drops them, and four values on the sheet — the recipient, the name after “Hi”, and two addresses — are covered by black bars, none of them supplied here. https://epstein-data.com/EFTA02078948 p.1. ↩

25. Gino Yu to Paul Schulte of Schulte Research, carrying the line “Can I forward this to my benefactor in NYC? You interested in meeting him while you are there?”, quoted inside a thread forwarded to Epstein. Both copies were read at the image on 2026-09-14, and both print a doubled space after “NYC?”, which the quotation above now preserves and an earlier version collapsed. https://epstein-data.com/HOUSE_OVERSIGHT_026036 p.1 is the first sheet of the forward, headed From: Gino Yu / Sent: 12/17/2016 2:43:49 AM / To: jeffrey E. / Subject: Fwd: Schulte Research: Deep dives into PRC fin tech ownership and Trump cabinet ownership. (DO NOT FORWARD), with the quoted line standing under On 17 Dec 2016, at 3:09 AM, Gino Yu … wrote:. A second copy is at https://epstein-data.com/HOUSE_OVERSIGHT_026042 p.1. An earlier version of this footnote said the sheet “could not be rendered”. The site’s /pdf/ store does return HTTP 403 for HOUSE_OVERSIGHT_* identifiers — a fact about that route — and the sheets are ordinary page images that read. ↩

26. Gino Yu, “Sent: Tuesday, August 30, 2016 11:26 AM”, subject “Intros for meeting on Sept 6th”, addressed to Epstein and to Andrew Link, opening “Hi Max, I’ve cc’d my benefactor, Jeffrey Epstein, to this mail.” Text layer only. https://epstein-data.com/EFTA00819732 p.1. Further copies at https://epstein-data.com/EFTA00819734 p.1 and https://epstein-data.com/EFTA01781890 p.1. ↩

27. Giuseppe Bersani to Gino Yu and Epstein, “Date: Tue, 21 Feb 2017 09:57:14+0000”, subject “Re: Any news?”, carrying Yu’s message of 21 February 2017 16:56 (“I’ve cc’d my benefactor, Jeffrey Epstein. He can make arrangements on the transfer of funds for the Conference.”) and his earlier message of 12:37 the same day. Text layer only. https://epstein-data.com/EFTA00692098 p.1. ↩

28. Lesley Groff to Nouriel, subject “Jeffrey Epstein”, “Date: Tue, 22 May 2018 16:23:00 +0000”, body “Hello Nouriel. I am Jeffrey Epstein’s assistant…Gino Yu suggested the two of you meet up in NY.”, https://epstein-data.com/EFTA00475071 p.1. Read at the image at 500 dpi by this pass on 2026-09-12: the sheet prints the date line “Tue, 22 May 2018”, where the text layer of the same sheet reads “The, 22 May 2018”, and prints an ellipsis character between “assistant” and “Gino”. Groff’s address and the recipient value are barred. A copy inside a longer thread is at https://epstein-data.com/EFTA00474163 p.2; the schedule entry “4:00pm Appt w/Nouriel (friend of Gino Yu)” at https://epstein-data.com/EFTA00475940 p.1; “8:00pm Appt w/Stuart Hameroff (friend of Gino Yu)” on an October 2016 schedule at https://epstein-data.com/HOUSE_OVERSIGHT_002356 p.1; “6/6:30pm Appt w/Ravi Mantha (friend of Gino’s)” for Thursday 12 January 2017 at https://epstein-data.com/EFTA02389108 p.1 and https://epstein-data.com/EFTA00443157 p.1; “Reminder: Peter Fenwick and Ray (Gino’s people) were asking if they could come see you today” for Monday 3 October 2016 at https://epstein-data.com/EFTA00317801 p.1 and https://epstein-data.com/EFTA01742498 p.1. All six of those text layer only; the House Oversight sheet could not be rendered, that store returning HTTP 403. ↩

29. Lydia Laurenson to Epstein, “Sent: Saturday, October 21, 2017 4:50 PM”, subject “Re: Intros”, carrying her earlier message (“Happy to meet up — any friend of Gino’s, etc.”) and Yu’s message of that morning introducing her. Text layer only. https://epstein-data.com/EFTA02376421 p.1. Peter Fenwick and Ray F. Lee are introduced in a thread opened by Yu on “14 September 2016 16:19”, https://epstein-data.com/EFTA01773706 p.2, text layer only. ↩

30. Gino Yu to Epstein, subject “Kuhn Interview… Divine intervention.”, “Sent: Wednesday, August 17, 2016 4:49 PM”; the sender value on this sheet is rendered “Gino V” and the body is one link and nothing else. https://epstein-data.com/EFTA02455066 p.1. Epstein to Gino Yu, “Date: Thu, 18 Aug 2016 13:58:06 +0000”, body “can you introduce me to kuhn by skype. I ve invited tagg to palm beach for lunch tomrow”, https://epstein-data.com/EFTA00821182 p.1. Robert Lawrence Kuhn to Epstein and Gino Yu, subject “Jeffrey - Robert”, “Date: Sun, 21 Aug 2016 02:30:55 +0000”, opening “Hi Gino: Appreciate the intro.” and carrying Yu’s message of “Aug 20, 2016, at 12:59 PM” and two earlier exchanges of 4 August 2016, https://epstein-data.com/EFTA00820611 p.1, a 5-sheet run. That third sheet was read at the image at 500 dpi by this pass on 2026-09-12, because the quotation from it turns on one character: the sheet prints “He has an interest in AI and consciousness”, and the text layer of the same sheet reads “an interest in Al” with a lower-case L. Kuhn’s and Yu’s addresses are barred there. The first two are text layer only for this page; the first and third were also read at the image by the Robert Kuhn pass and are cited there. ↩

31. Gino Yu to Epstein, “Subject: Re:”, “Date: Sun, 07 Aug 2016 13:20:15 +0000”, carrying Epstein’s message of “7 Aug 2016, at 8:40 AM” (“why dont we just put david to his own claim.”) and Yu’s own of 8:36 AM. Read at the image at 500 dpi by this pass on 2026-09-12 (and at 300 dpi by the writer). His address is covered by a black bar at both points it appears, and the paragraph beginning “Whatever the costs to set up” carries two black bars, one at the end of its first sentence and one at the start of its second; the quotation in the body now runs to the end of that paragraph so that both appear, and nothing is supplied in their place. An earlier draft stopped the quotation at the first bar, which left a reader counting bars one short of the footnote. The paragraph the body quotes second is the one the sheet opens “PPS.”, and a postscript about the Shanghai conference stands between the two. https://epstein-data.com/EFTA00822391 p.1. A second copy of the same message is at https://epstein-data.com/EFTA02455131 p.1, text layer only. Epstein’s own message of that morning stands alone at https://epstein-data.com/EFTA02455029 p.1. ↩ ↩² ↩³

32. Gino Yu to Epstein, “let me know what conditions you have on the remainder of the 100K”, https://epstein-data.com/EFTA00940050 p.1, with a second copy at https://epstein-data.com/EFTA02557855 p.1; and “Could also use the remainder of the 100K funds that you’ve offered me a while back”, https://epstein-data.com/EFTA00906974 p.1. Both text layer only. ↩

33. Gino Yu to Epstein, “Subject: Re:”, “Date: Sat, 25 Nov 2017 00:38:42 +0000”, the message that accounts for the 100K, describes the Havana meeting, names the “stage 3” model and closes on the funding he had obtained elsewhere. Text layer only. https://epstein-data.com/EFTA00984125 p.1. A second copy is at https://epstein-data.com/EFTA02581361 p.1. ↩ ↩²

34. Giuseppe Bersani to Epstein, attaching “Report_on_Neupsyco_Congress.pdf”, “Dear Jeffrey, I enclose just a brief report of the Neupsyco Congress in La Habana on November 6-8. No words enough to express gratitude for your enormous support, that made the event possible.” Text layer only. https://epstein-data.com/EFTA00941140 p.1. A second copy is at https://epstein-data.com/EFTA02559288 p.1, and a further message of thanks after the event at https://epstein-data.com/EFTA00957399 p.1. ↩

35. Stuart Hameroff to Epstein, “Subject: Re:”, “Date: Thu, 30 Mar 2017 21:17:39 +0000”, attaching “RPISanDiego.docx”, answering Epstein’s question of that afternoon, “did it get resolved. . what is your side”. The passages quoted above stand on printed p.1 and the closing “Wanna help?” on printed p.2. Text layer only. https://epstein-data.com/EFTA00643937 p.1; https://epstein-data.com/EFTA00643937 p.2. Because this footnote names both sheets of a two-sheet run, a mechanical test of its pins cannot fail them; the split is stated here so a reader can check it by eye. Epstein’s reply of the next day, “who are the speakers and what is the cost. Im in for at least 50k, before knowing anyhing”, is at https://epstein-data.com/EFTA01050783 p.1, text layer only. Yu’s own announcement of the Shanghai meeting is at ³¹. ↩

36. Peter Fenwick to Epstein, “Sent: Friday, November 24, 2017 6:26 PM”, subject “RE: Payment for research”, body “Entirely agree, I have dropped Gino from the =eam and will make sure that I control the =thers. / Peter”, carrying Epstein’s message of 24 November 2017 beneath it (“and unless you take control I must stop =ny serious consideration of continued support.”). Read at the image at 600 dpi by this pass on 2026-09-12: the = characters are PRINTED ON THE SHEET. They are quoted-printable soft breaks carried onto the face by the export, not extraction damage, and this desk’s ruling is that they are never repaired. An earlier draft of this page printed “team”, “others” and “any”, and said in this footnote that the sheet “carries transport damage and both quoted spans print undamaged” — a statement about an image on a sheet the same footnote declared it had not opened. It is false: the damage is inside both quoted spans. FTS5 returns no other copy of either sentence in the release, so there is no clean copy to prefer. Fenwick’s address is barred at three points. https://epstein-data.com/EFTA02581044 p.1. ↩

37. Epstein to Peter Fenwick, “Sent: Saturday, August 11, 2018 2:37 PM”, subject “Re: Final test for effect”, body “reminder when i asked gino for his breakdown was when i fo=nd out 50% went for drinks and food”, carrying Fenwick’s message of that morning and Epstein’s refusal of 9 August (“No . Sorry have yet t= see any indication of something meaningful”). Read at the image at 600 dpi by this pass on 2026-09-12: the = characters are PRINTED ON THE SHEET, in both quoted spans, as they are throughout this export (“f=scinating”, “sen=ations”, “answ=r”, “eno=gh”). An earlier draft printed “found” and “to see” and said in this footnote that the quoted spans print undamaged; they do not, and the words are not supplied. FTS5 returns no other copy of either sentence. Fenwick’s address is barred at five points, all five counted at a crop of the 600 dpi render: the “On Sat, Aug 11” attribution line and its <mailto:> continuation, two on the quoted To: P=ter Fenwick <…> <mailto:…> line, and the “On Thu, Aug 9” attribution line. An earlier draft of this footnote said four. The sheet’s own top-level To: value, “Peter Fenwick”, is not barred at all; what is barred is the address every time. https://epstein-data.com/EFTA02603273 p.1. ↩

38. Lesley Groff to Epstein, subject “Snacks at 1:30meeting with Gino and Mitch?”, https://epstein-data.com/EFTA02601405 p.1; Epstein’s answer, “Sent Tue 7/17/201811:37:11 AM”, “no. just some cookies.”, https://epstein-data.com/EFTA02251117 p.1; the office message to Yu of “Mon 7/16/2018 3:00:26 PM” confirming “you will come see Jeffrey tomorrow”, https://epstein-data.com/EFTA02252343 p.1; and the alert “Gino needs to confirm 2:30pm tomorrow Mitch come?”, https://epstein-data.com/EFTA02252167 p.1. All text layer only. ↩

39. Gino Yu, “Sent: Thursday, September 13, 2018 8:47 AM”, subject “Intros”, body opening “Hi Nathalie, Jeffrey can meet Friday or Saturday” and closing “Greetings from Zurich. Here to sign the space in Davos.” Read at the image at 400 dpi by this pass on 2026-09-12. The sheet carries quoted-printable damage in the sentence that follows the opening (“a spec=fic time”) and both quoted spans print undamaged, which this pass established at the image. Yu’s address and one recipient value are barred. https://epstein-data.com/EFTA02623145 p.1. ↩

40. Joscha Bach to Lesley Groff, “Date: Sun, 20 Jan 2019 19:12:10 +0000”, subject “Fwd: davos accomodation”, attaching “Klosters_Housing_Guanxi_Joscha.pdf”, https://epstein-data.com/EFTA00489837 p.1; Bella Klein to Richard Kahn and the office, “Sent Mon 1/28/2019 9:36:35 PM”, asking “Did we received invoice for accommodation”, https://epstein-data.com/EFTA02274006 p.1; the answer “Rich received the invoice for accommodation. We also purchased a round trip flight to Zurich for Joscha. I am not sure what entity is paying.”, https://epstein-data.com/EFTA02274033 p.1; and Klein’s reply of “Jan 28, 2019, at 4:47 PM”, “ok, its paid already.”, https://epstein-data.com/EFTA02274092 p.1. All text layer only. The chalet arrangements, in which Bach writes “I assume this is the place that we share with Gino et al?”, are at https://epstein-data.com/EFTA02276776 p.1. ↩

41. Gino Yu to Joscha Bach, copy to Lesley Groff, subject “Re: Contact”, “Date: Fri, 11 Jan 2019 05:03:40 +0000”, body “I will also have a shaman with me as well as a psychonaut that grows his own mushrooms.”, carrying Bach’s message of “11 Jan 2019, at 12:56 PM” beneath it. Read at the image at 300 dpi by the writer on 2026-09-12. All three correspondents’ addresses are covered by black bars, and one further value inside the quoted message is covered by a black bar; nothing is supplied here in its place. https://epstein-data.com/EFTA00488492 p.1. ↩

42. The conduit, followed. http://hinmancharity.org/about/ — the address Yu gave Epstein, transcribed from the image of https://epstein-data.com/EFTA00863609 p.1 — was live and returned a page when this pass read it on 2026-09-12. It presents the site as “The Hinman Charity”, carries the line “Preserving Mystical Cultures”, and states that “The Charity was established in 2013 and blossomed out of the Hinman Foundation which was established in 2004” and that it is guided by Buddhist principles and generally supports Buddhist communities. This is an outside source, not a document in the release, and it is cited only for what the organisation says about itself. Against the corpus full-text index on the same day, FTS5 exact phrases: hinmancharity 2 documents, https://epstein-data.com/EFTA00863609 p.1 and https://epstein-data.com/EFTA02509724 p.1, which are two copies of the same message; "Hinman" 53 documents corpus-wide, 0 inside this population — they are a securities-regulation thread, a law-firm partner list, an architect and a gallery contact, and the tokeniser does not reach the two above because the extraction runs the word into the domain; "Hinman Charity", "Hinman Foundation" and "Mystical Cultures" 0 each. The name of the trustee the site’s About page gives was also run as an exact phrase and returns 0; it is not printed on this page. This is recorded as a closed negative: the offer is in the release, the organisation is real, and the release shows nothing following from either. ↩

43. The terminus test. Each of the twelve forms named in the body was run as an exact FTS5 phrase against the whole index and intersected with this population. Eleven intersections are empty. "refund" is not: it returns 9,742 documents corpus-wide and three of them are inside this population, two in the 1,086-document base phrase. They are https://epstein-data.com/EFTA00678099 p.2 and https://epstein-data.com/EFTA01050786 p.2, two copies of the Center for Consciousness Studies notice moving TSC 2017 from Shanghai to San Diego (“In some cases, Crystal Globe has refunded money, but in other cases they haven’t, and have continued to take more. CCS will reimburse our registrants”), which is the episode set out under Shanghai, and the conference that came apart; and https://epstein-data.com/EFTA02563991 p.3, a course-provider exchange about a refundable book deposit (“please confirm that i should pay the refundable book deposit”), which is in this population only because a different message earlier in the same run, at https://epstein-data.com/EFTA02563991 p.1, reads “Last night we met with one of your scientists, Gino”. All three text layer only. Because this footnote names two sheets of one run, a mechanical test of its pins cannot fail them; the split is stated here so a reader can check it by eye. The other eleven forms were re-run at the same time and return nothing inside the population; "repay", "repaid", "returning the" and "terminate the" do return documents inside the 359 forename residue, but every one of those is in a bucket assigned to somebody else or to noise. The year tally is taken from the first fourteen non-blank lines of each document’s own first sheet and reads 2015: 192, 2016: 462, 2017: 367, 2018: 187, 2019: 45, with 18 earlier documents (one each in 2001, 2008 and 2010, two in 2011, eleven in 2013 and two in 2014) and 31 yielding no year in that window. 192 + 462 + 367 + 187 + 45 + 18 + 31 = 1,302. A test that reads only the head of the first sheet cannot see a date that stands on sheet three, and that is the limit of this measurement. ↩