Epstein Data Wiki

Jeremy Rubin

Properties2
tagsperson, bitcoin, cryptocurrency, massachusetts-institute-of-technology, mit-media-lab, digital-currency-initiative, gratitude-america, venture-capital, tax-structuring
aliasesRubin, Jeremy, jeremy rubin, @JeremyRubin

67 min read

How he appears in these documentsWriting to Jeffrey Epstein in his own voice under the sender name “Jeremy Rubin”, signing “Jeremy” and, on many sheets, over the printed handle “@JeremyRubin”123
Described in the released files as”Jeremy Rubin, MIT student who started the MIT bit coin p=oject”, in a correspondent’s message to Epstein of 21 June 2014 whose sender name is barred4
What Joi Ito called him”the kid that I was also talking about in the context =f the MIT Bitcoin project”5
What Epstein’s office called him”Jeremy Rubin and the Bitcoin troops”, on a tentative Cambridge day sheet of 9 October 20156
How the correspondence beganAn introduction of 25 June 2014 from a correspondent whose name and address are barred on the sheet, to “Jeremy Rubin” and “Jeffrey Epstein”, copying Lesley Groff and Joichi Ito, subject “Introduction”7
Span of the documents read for this page21 June 2014 to 8 November 201848
What he asked forOn 27 December 2015: “I was wondering if you would b= interested in financing my continued research in this space”. Epstein answered on the same day’s headers: “yes , we can do it . lets think of best =tructure. put together a rough budget.”19
What the office papered it asA “consulting agreement” sent to him by Richard Kahn with a cheque, in early February 201610
What was paid, as read at the cheque imagesTwo Gratitude America, Ltd cheques on Deutsche Bank Trust Company Americas, each $25,000: no. 362 of 4/19/2016, memo “2nd quarterly payment”, and no. 365 of 8/1/2016, memo “3rd Quarterly Payment of 4”. The other payments in the schedule are attested only in correspondence and no cheque for them was read for this page111213
The 2018 fund”Deploy Capital”. Epstein pressed for a larger position in its first deal, told Rubin “dont worry about the size of fund”, and was asked by Rubin “would you put more money into the fund if I am bringing in other good deals in the immediate future?” The only statement on these sheets about the fund’s own size is Rubin’s2
16 to 19 July 2018Rubin proposed a form of words to keep Epstein’s role in the fund from investee companies, under the subject “Fund Optics”; Epstein replied “you are right re appearance”; Rubin drafted a telephone script under which Epstein would speak to a target company under a first name only; Epstein answered “phone fine.”, then “thurs”, then “you should be on”314151617
What he has said publiclyDL News and the Marblehead Current reported that on 2 February 2026 he posted on X that he was “glad the emails are being released”, that he had “some professional engagement” with Epstein which was “never exclusive”, and that “I hope the release of the emails brings us closer to justice for those harmed and a better understanding of the nature of corruption in our society”1819
What no document read for this page showsAny appearance by him on a flight log, a passenger manifest or a property record, or at Little St. James; any contact after 8 November 2018; any document pairing his name with the vocabulary of the offence20

Jeremy Rubin is a Bitcoin developer. At the start of this correspondence he was an undergraduate at the Massachusetts Institute of Technology who had founded the MIT Bitcoin Project; the documents read for this page later place him in San Francisco, working on Bitcoin protocol development and, from 2018, running a venture fund.

Documents released by the Justice Department under the Epstein Files Transparency Act, together with four sheets from the House Oversight production, record two-way correspondence between Rubin and Jeffrey Epstein from June 2014 to November 2018. Over that period Epstein agreed to finance his research, his office papered the arrangement as a consulting agreement and paid it in quarterly cheques drawn on Epstein’s foundation, Rubin introduced at least four people to Epstein and carried investment proposals to him, and by 2018 a venture vehicle called Deploy Capital was running, with Rubin choosing the deals, Epstein setting the size of the position to be taken in the first of them, and Epstein writing of himself that he was “more than happy to fund things”.

On 16 July 2018 Rubin put to Epstein a plan for describing the fund to prospective investee companies without naming him, and Epstein approved it. The next day Rubin drafted a script under which Epstein would join a due diligence call under a pseudonym, and Epstein accepted, fixed the day, and told Rubin to be on the call as well. Both sides of that exchange survive in the release, on separately produced sheets, in the correspondents’ own words.

On the morning of the call, two days later, one correspondent inside Epstein’s office instructed another to dial him in and not to say his surname. That instruction is the office’s own and Rubin is not a party to it, and it is set out below under a heading that says so. What any of this amounts to is not for this page to decide; the documents are set out below and quoted in full at the points that matter.

This page makes no allegation of wrongdoing against him. Nothing in the documents read for it suggests any knowledge of or involvement in the conduct for which Epstein was charged, and no document read for this page names him in connection with it. He is a living adult, named in the documents in the clear and named here in full. A public statement by him about the release exists and is reported below; no response by him to any individual document on this page appears in the release or in any source read for it.

Three things to get right before reading further

First, the same message exists in copies that differ on the paper. Much of this correspondence was produced twice, once as an Outlook-style print and once inside a later Gmail-style thread. The two do not always print the same characters. Rubin’s message of 16 July 2018 prints with quoted-printable corruption on its own sheet, differently corrupted inside Epstein’s reply, and cleanly inside the later Gmail-style thread; Epstein’s reply prints with corruption on its own sheet and cleanly inside that same later thread. Where a fault is on the paper it is reproduced here unrepaired, and each footnote says which copy is being quoted and whether it was read at the image or only in the text layer.31415

Second, the copies do not agree on the clock, and this page does not reconcile them. Rubin’s “Fund Optics” message carries the header time 6:29 AM on its own sheet and is quoted back at 2:29 AM inside the reply. Epstein’s reply carries 10:22 AM on its own sheet and is quoted back at 3:22 AM inside the later thread. The two gaps are not the same length. The same fault runs through the whole correspondence: on one sheet a message of Epstein’s stamped 2:23 PM answers a message of Rubin’s stamped 5:15 PM the same day, and on another a message of Epstein’s stamped 5:18 PM answers one of Rubin’s stamped 9:17 PM. No interval between any two messages on this page is computed across a header and a quote-back, and no time on this page is converted to any other zone.21

Third, the Outlook export sometimes drops a message body entirely. The sheet produced for Rubin’s introduction of Gerald Banks to Epstein prints the header block and the sign-off and nothing between them: at the image the space where the body belongs is blank, with no redaction bar on it. The body of that message survives on a separately produced sheet of Epstein’s forward of it, and it is that sheet this page quotes.2223

June 2014: “He’s open to meeting you”

The earliest document located for this page is of Saturday 21 June 2014, and it is one of two messages sent to Epstein that evening by a correspondent whose sender name is barred on both sheets. The first, at 9:46 p.m. by its header, is subject “At a session…”:

On alternative currencies at geek foo Reid and Joi are also at this.

At lunch I sat with Jeremy Rubin, MIT student who started the MIT bit coin p=oject. He’s wicked smart. He’s open to meeting you. worth it!

The closing name on that sheet is blacked out.4 Just under an hour later the same correspondent wrote again, subject “Also”: “Joi also sync’ing with Jeremy rubin. Very interesting guy!”24

On Wednesday 25 June 2014, at 00:07:41 +0000 by its own header, the introduction went out. The sender’s name and address are barred; the To line lists “Jeremy Rubin” and “Jeffrey Epstein”, the Cc line lists “Lesley Groff” and “Joichi Ito”, and the subject is “Introduction”. The body reads in full:

Hi Jeremy!

This is an introduction to Jeffrey, an NYC based financier who is a friend of Joi Ito and me.

He has an interest in alternative currencies and would like to organize time to Skype. His assistant, Lesley, is copied. You all can take it from here.

The message closes “Cheers,” and its signature block carries a value barred by the producing party.7

Joi Ito answered later that morning, copying Rubin, Epstein and Groff. His reply opens “Thanks Linda.” and continues: “Jeffrey, Jeremy is the kid that I was also talking about in the context =f the MIT Bitcoin project.”5 No document read for this page gives that correspondent’s surname, and none is supplied here. Her own messages carry her name barred on their face; the forename above is quoted from Ito’s message because that is where the sheets print it.

October 2015: “the Bitcoin troops”

On Friday 9 October 2015 at 10:18:17 PM, Lesley Groff sent Epstein a tentative schedule for a day in Cambridge the following week, headed “Fri. Oct. 16, 2015 NY-Cambridge” and marked Importance: High. The 1:30 p.m. line reads:

1:30pm TENTATIVE Appt w/Jeremy Rubin and the Bitcoin troops

The same sheet lists appointments with Joscha Bach, Neri Oxman and Martin Nowak’s group, and an evening entry at the Chomskys’ apartment. Four of the six entries on it, including the 1:30 p.m. line, are marked TENTATIVE; a 4:00 p.m. line about a seminar is unmarked and the last entry reads “TBD EVENING”. The covering message says confirmation of timing was still being sought from everyone. The sheet records a plan, not an occurrence.6

Twelve days after that, on 21 October 2015, Rubin made the first of the introductions set out below, of the cryptographer Madars Virza.25

December 2015: the ask, and the answer

On Sunday 27 December 2015, under the subject “Jeremy’s plans for next year:”, Rubin wrote to Epstein from Israel. The sheet carries quoted-printable damage on its face and is quoted here as it prints:

I’m trying to commit to plans for next y=ar and figure out what will be the best way for me to continue to learn, b=ild experience, and find innovative angles for cryptocurrency tech. <=div>

I’m currently in Israel with spotty wifi but I=wanted to reach out before the new year to start a discussion. To be direc= (and send a message before wifi cuts out), I was wondering if you would b= interested in financing my continued research in this space, or if there =re any projects you’d want to push forward that I might play a roll in= I’d also love to learn more from you about how financial markets real=y work and build some of my own “exploits” at some point. I= you think this could be fruitful, it’d be great to find a time to hop=on a call and try to iron out details.

1

Epstein’s answer is one line:

yes , we can do it . lets think of best =tructure. put together a rough budget.

9

The two Outlook headers are two minutes apart, Rubin’s stamped 1:42 PM and Epstein’s 1:44 PM. The quote-back inside Epstein’s reply attributes Rubin’s message to 9:41 AM instead. The headers are reported here as they print and no interval is taken across the two clocks.21

January to August 2016: the consulting agreement and the cheques

On 24 January 2016 at 8:15:08 PM EST, Epstein forwarded to Richard Kahn a message Rubin had sent him under the subject “Info:”, with a covering line of his own: “informat=on you need from me.” Rubin’s message reads: “Please let me=know the specifics I’ll need as well for taxes etc & any other”.10

What Kahn sent back is the nearest thing in the release to the terms of the arrangement, and it is quoted inside the same thread:

Jeremy,

I a=ologize for the delay in processing your check and consulting paperwork.&nb=p; I will be sending today via Fed Ex to your Cambridge address a check in t=e amount of 25,000 along with 2 copies of consulting agreement.

P=ease sign one copy and return to address below and keep the other copy for y=ur records.

Rubin acknowledged receipt on 4 February 2016 at 4:52 PM: “I have recei=ed the Fedex, and will review and send the documents shortly.” He noted a typographical error in his own email address on the document and asked whether to strike it or be sent a fresh copy; Kahn answered three minutes later on the same clock, “Ok to just manually change. Thank you.” Twelve days after that Rubin asked “Did you receive the p=perwork today?”10

The consulting agreement itself was not located by the searches described at the end of this page, and the sheets quoted above are the only description of its contents read here. They do not state its term, its scope of work, or the total consideration.

Two cheques to Rubin were read at the image for this page. Both are drawn on Gratitude America, Ltd at Deutsche Bank Trust Company Americas, New York, both are for $25,000, and both carry a handwritten authorised signature which this page does not identify.

ChequeDate on its faceAmountMemo
3624/19/2016$25,000.00”2nd quarterly payment”11
3658/1/2016$25,000.00”3rd Quarterly Payment of 4”12

The memo on cheque 365 states the schedule: four payments. No cheque for the first or the fourth payment in that schedule was located by the searches described at the end of this page. This page therefore does not state a total. The existence of further payments rests on correspondence, set out below, and on nothing read at a cheque.

Each cheque prints a payee address block beneath the payee name. On cheque 362 the street line is blacked out and the town line reads Cambridge, Massachusetts; on cheque 365 the street line is blacked out and the town line reads Marblehead, Massachusetts. Neither street line is reproduced here and neither is reconstructed.1112

The date the office and the cheque do not agree on

On Thursday 7 July 2016 Richard Kahn wrote to Epstein under the subject “jeremy rubin / leon botstein”, correcting a report he had given him earlier. The sheet carries damage on its face:

i apologize as i gave you bad information regarding jeremy =ubin and leon botstein jeanne and una stated in emails =elow that 5-1-2016 checks were outstanding as of 5-30-2016 and =-27-2016 which was inaccurate as they both were cashed i made the mistake of relying on their emails below and not =hecking myself

both Botstein and Rubin have received and cashed their =-1-2016 and 5-1-2016 checks next checks due for =oth is 8-1-2016

13

The first of those two dates is damaged on this sheet. A separately produced copy, inside Epstein’s own forward of the message six minutes later by the Outlook headers, prints it as “2-1-2016”: “both Botstein and Rubin hav= received and cashed their 2-1-2016 and 5-1-2016 checks next che=ks due for both is 8-1-2016”.26 Quoted inside both copies is an office message of 27 June 2016 listing two outstanding cheques, “Jeremy =ubin” at $25,000.00 and “Leon =otstein” at $37,500.00, and an earlier one of 14 June reporting that Gratitude cheques to Botstein and Rubin, “dated 5/1/2016”, were “=utstanding as of 5.30.2016” and had “=ot cleared as of today”.13

⚠ Those three office messages describe the second payment as a cheque dated 5/1/2016. The cheque in hand for the second quarterly payment is dated 4/19/2016 on its own face. The documents disagree and this page does not resolve them. It records only that the office’s running description of the schedule (2-1-2016, 5-1-2016, next due 8-1-2016) and the dates printed on the two cheques read here (4/19/2016 and 8/1/2016) do not line up on the middle payment. The first payment is not in hand either, and the desk went looking for it. The bank production’s Gratitude America cheque images, tabulated on the Gratitude America page, carry no February 2016 cheque to anyone; the corpus returns no memo reading “1st quarterly payment”; and the only sheet in the release bearing the memo form used on the August cheque, besides that cheque’s two copies, is a cheque to a different payee. So a third date now stands against the other two without a document behind it: an outside report of February 2026 gives the first payment as 3 February 2016, where the office’s own running description gives 2-1-2016. This page adopts neither. It records that the first cheque’s image is not in the release, that two sources disagree about its date by two days, and that nothing read here settles which is right.27

Epstein forwarded Kahn’s correction to Karyna Shuliak the same afternoon with a five-word covering note: “see what happens when no one punished”.26 The note comments on his own office’s reporting error; nothing on that sheet refers it to Rubin.

Four months earlier, on 8 March 2016, Kahn had chased Rubin directly: “hope you are well wanted to=make sure you have check from gratitude as my bookkeeper just informed me t=at your check has not cleared please advise”. Rubin replied the same evening: “Sorry, de=ay is my fault, will clear it by end of week.”10

The introductions

Rubin wrote to Epstein and a third party jointly on at least four occasions, putting the two of them on the same To line and describing each to the other. Two of the four carry a subject line beginning “Intro”; the other two do not.

21 October 2015, Madars Virza. To “Jeffrey Epstein; Madars Virza”, copying “Joichi Ito”, subject “Skype with Madars:”: “I wanted to introduce you to Madars,=one of the most brilliant cryptographers I know.” The sheet carries transmission damage throughout and is quoted unrepaired.25 The correspondence that followed is set out on Virza’s page.

12 November 2015, Bryan Bishop. To a To line listing “Jeffrey Epstein; Joichi Ito; Bryan Bishop”, subject “CFTC to regulate bitcoin”: “Bryan (CC’d) is doing som= work looking at how cftc is trying to regulate bitcoin options trading.”28 The correspondence that followed is set out on Bishop’s page.

16 June 2016, Suhas Daftuar. To “Jeffrey Epstein; Suhas Daftuar”, subject “Intro:”:

= wanted to introduce you guys.

Suhas is one of the founders of Hudso= River Trading and is now running a Bitcoin R&D lab called Chaincode a=ong with one of the other HRT founders, Alex Morcos.

Jeffrey is very=keen on the cryptocurrency space; in particular he has a unique perspectiv= on the regulatory/tax side of things.

Epstein answered three minutes later on the Outlook headers, copying Daftuar: “Thx look forward to it”.2930

25 January 2018, Gerald Banks. To “Gerald Banks; jeffrey E.”, subject “Intro Gerald <> Jeffrey”. The Outlook sheet prints no body at all. The body survives inside Epstein’s forward of the message to Lesley Groff the same afternoon.22 On that second sheet it reads, at the image:

As noted, I wanted to intro you to Gerald Banks at Altpoint Capital. I’m working with Gerald on the crypto volatility fund and exchange project I mentioned.

Perhaps you can arrange to meet in NYC next week to discuss and learn more.

23

Epstein’s office arranged the meeting for the following Monday, 29 January, and its day sheet records the appointment in the standing formula “Appt w/Gerald Banks w/Altpoint Capital (friend of Jeremy Rubin)”; that sequence is set out on another page. On 5 February 2018 Rubin asked Epstein “How was the meeting?” and Epstein answered the next day. The reply sheet prints, beneath Epstein’s own header, “I liked him” and then “How was the meeting with Gerald?”, with Rubin’s “Curious for your thought.” beneath and no attribution line of its own printed above it. The sheet does not identify whom “him” refers to and this page does not supply it.3132

A fifth sequence belongs beside these and is not an introduction of the same kind. In March 2016 Bryan Bishop introduced Rubin to LedgerX’s chief technology officer Zach Dexter; Rubin met the company, forwarded its deck to Epstein under the subject “[Potential Investment] Fwd: LedgerX deck”, visited its offices and reported back. That is Rubin carrying a company and its material to Epstein, not an introduction of a person to Epstein under a joint To line, and it is described as such here. The sequence is set out on Bishop’s page.33

What he brought Epstein: structures and tax

Alongside the introductions, Rubin repeatedly put structuring questions to Epstein. Two are worth quoting because they show the register of the exchange.

On Friday 30 September 2016, subject “Housing Idea”, he set out a scheme for deducting New York rent through a corporation and a bond, worked it through seven numbered months, and closed:

Of course, you should (probably) make these bonds mature on a 1yr period to get long-term rates, but what’s the bug in this? Does this work? Can I do it to save taxes/rent? Is the part where you get a preferential rate rent as a shareholder illegal? How many days out of the year can you stay there before it become not short term housing, and you can no longer deduct your rent?

34

On Friday 1 September 2017, subject “Question”, he set out a scheme for holding Bitcoin inside a self-directed retirement account through a wholly owned company, and asked: “Am I missing a step here or is this an interesting new technique?”35 Epstein answered within the hour on the produced headers with two messages, the second of which reads: “also bitcoin is currently not a currency ,, but property , each coin has its own cost basis…”36

December 2016: renewing the contract

On Thursday 8 December 2016, under the subject “Renewing Contract for 2017”, Rubin wrote:

Wanted to discuss renewing my contract for 2017. I’m trying to plan my financials to a certain degree for the next year before this tax year closes.

Having a regular source of income with loose requirements has been very useful for me to be able to apportion my time into deep thought in the space without worrying about immediate returns. But it would also be good, perhaps, to discuss if I’m meeting your expectations, and what topics you’re interested in through 2017 / if you’d want to see more outputs from me directly (some kind of quarterly report/meeting or something?).

He then listed what he had spent the previous month on, including “doing research on some new smart contracting primitive constructions”, “attempting to mathematically formalize parts of bitcoin programs” and “continuing my work on open source bitcoin software”. Epstein’s replies in the same thread are logistical: “uncertain. until jan”, “im in new york all next week, cambridge on sunday”, “skype ok, maybe over the weekend”.37

No document read for this page records the outcome of that request: no renewed agreement, no 2017 cheque, and no message in which either man refers back to it.

2017: a gap in the production

Of the 726 documents returned by the exact-phrase query "Jeremy Rubin" against the release’s full-text index, the year of the first header date printed on page 1 falls as follows.

YearDocuments
201463
201599
2016342
201716
2018175
no header date on page 131

Those sixteen 2017 documents are copies of two email exchanges and nothing else: a thread of 1 and 4 February 2017 under the subject “Greetings from Japan”, and the “Question” exchange of 1 September 2017 quoted above. Eight documents each.38

That is recorded here as a gap in the production and is not explained. Both sides’ correspondence on either side of it presupposes a live relationship: in December 2016 Rubin was asking to renew a contract for 2017 and describing a quarterly reporting cadence, and by early 2018 he was introducing people to Epstein and, by the middle of that year, the fund described in the next section was operating. This page does not assert that the 2017 correspondence was withheld, and it does not assert that it never existed. It records that the release, searched as described at the end of this page, yields two exchanges for that year against 342 documents for the year before it.

2018: Deploy Capital

By the middle of 2018 the relationship had a vehicle. On Monday 25 June 2018 Rubin wrote to Epstein, copying Richard Kahn, under the subject “Exchange/OTC Accounts Set Up”:

To set up the exchange and OTC =ccounts I need the following info:

  • Busi=ess Address
  • Fund Administrator/Account Representative ->= is this me, or do we have some sort of ‘registered agent’?

I’ll also need to know bank account informat=on to be able to wire money over.

Epstein’s reply that morning was “good think we didnt buy coins at 6700”; Rubin answered “I’ll get some cash deposited on the exchanges so that we can =ull the trigger at the right time.”39

The fund’s first proposed deal was a company called Layer One. On Saturday 14 July 2018 Rubin sent Epstein a structured memorandum on it. The figures on that sheet were read at the image, because the text layer misreads two of them:

Proposed Investment: $250k for 1.7% of the company. I’d be propose a larger investment if you want to invest more aggressively, but I don’t think I would want to invest more than 10% of Deploy Capital’s funds (500k for 3.5%) so we have space for other deals I’m looking at, but I’d like your guidance on this.

2

Epstein answered by raising the number. Across that Saturday, in a thread whose quoted attributions run on more than one clock, he wrote “if you like them lets talk / meet them. they want to raise 20m?? for a carried interest fund? or build a co?”, then “send papers. I think a larger position. 2 mil for 20 % . or 2 mill for 15 plus warrants. ?”, and, when Rubin hesitated, “dont worry about the size of fund . concentrage on what you think are the best people and futures. papers mean investment docs”. Rubin answered with a question about who would put more in:

I’m also slightly reluctant to do 2 mil for 20 because:

  1. there are some other interesting projects I’m looking at, so am not sure it makes sense to put 40% of the fund in — would you put more money into the fund if I am bringing in other good deals in the immediate future?

2

Two days after that memorandum, Rubin raised the subject of how the fund would be described.

16 to 19 July 2018: “Fund Optics”

On Monday 16 July 2018, under the subject “Fund Optics”, Rubin wrote to Epstein. The sheet produced for that message is an Outlook print and it carries quoted-printable corruption on the paper, including two lines of markup printed as text. It is quoted here exactly as it prints:

Hi Jeffrey,

It’= probably best for me to just stick to saying “I raised a fund” =oing forward (I haven’t said anything inconsistent with this). I’m=concerned otherwise potential investments googling your name might get spo=ked (especially given SF politics). =br> I talked with Vincenzo about thi= and we agreed the above line would be inconspicuous enough.

If you =o want to talk to a potential investment directly, I can just say that you=are a strategic LP and you wanted to talk, which should dilute their conce=n and explain why you wanted to talk.

I’d like your guidance=& approval in case Vincenzo and I missed something with this plan. =/div>

Jeremy

3

A second copy of the same message, quoted back inside a later sheet of the same thread, prints it clean. In that copy the four paragraphs of the message read: “It’s probably best for me to just stick to saying “I raised a fund” going forward (I haven’t said anything inconsistent with this). I’m concerned otherwise potential investments googling your name might get spooked (especially given SF politics).” / “I talked with Vincenzo about this and we agreed the above line would be inconspicuous enough.” / “If you do want to talk to a potential investment directly, I can just say that you are a strategic LP and you wanted to talk, which should dilute their concern and explain why you wanted to talk.” / “I’d like your guidance & approval in case Vincenzo and I missed something with this plan.”15 That the same production renders the same message two different ways is a fact about the release and is recorded rather than smoothed.

Epstein replied the same morning. His own Outlook sheet is also damaged on the paper, and prints:

you are right re appearance. however we =houdl find a way for me to speak to the team maybe skype. investments are people focused. and neither you nor v=ncenzo yet have the experience. . anyway good call=

14

The clean copy of the same reply, quoted back inside the later thread, reads:

you are right re appearance. however we shoudl find a way for me to speak to the team maybe skype. investments are people focused. and neither you nor vincenzo yet have the experience. . anyway good call

15

“shoudl” is the writer’s own transposition and is not mended on either copy. On the damaged sheet the initial “s” is additionally eaten by the transmission fault, so that sheet prints “=houdl”.

On Tuesday 17 July, Rubin proposed the mechanism. The message is clean at the image on the thread sheet:

How would you feel about this:

I’ll just schedule a time for them to talk to my “due diligence advisor/team”. You can phone in and ask whatever you like, under a pseudonym (even just first name, not last name). You can open with something like, “Hi this is Jeff, I’m helping Jeremy check out what you’re planning to do with Layer One and make sure it’s going to be a good investment for Deploy Capital. Can I dive right in to questions?”

I think that should be fine, and is not inaccurate, just imprecise.

If you did want to video them it’s a bit more complicated, but I can come up with something.

Epstein’s answer on the same sheet is two words: “phone fine.”15

Rubin offered times (“Thursday or Friday this week between 12:30-2:30pm ET work for you?”) and gave the telephone numbers of two people at the company, printed on the sheet by forename only with the numbers themselves blacked out. He also asked “Did you want me on the call as well?” Epstein wrote “confernece call line pref”, and Rubin answered on Wednesday 18 July: “Ok I’ll set one up. / Which time did you want to pick?” Epstein picked a day in one word: “thurs”.1516

The thread does not end there, and the last word in it is Epstein’s. In the small hours of Thursday 19 July Rubin sent dial-in details under the heading “Call Info:”, with the line “Call starts at 12:30 ET”. The details themselves are blacked out and are not reproduced. He asked one more question:

By the way, did you want me to be on the call or just one on one with them?

16

At 09:32:49 +0000 that morning Epstein answered, in four words:

you should be on

17

Set out as documented fact, and no further: Rubin proposed a form of words for describing the fund that omitted Epstein, asked for his “guidance & approval”, and recorded that a third person had agreed the wording; Epstein answered that Rubin was “right re appearance” and asked for a way to speak to the target company himself; Rubin then drafted a form of introduction for Epstein to use on that call which gave a first name and no surname and described him as part of a “due diligence advisor/team”; Epstein accepted, fixed the day, and told Rubin to be on the call as well. Each of those steps is quoted above from a primary sheet, in the writer’s own words, with both sides of the exchange present. This page states what the documents say and stops.

What Epstein’s office did on 19 July, which is not Rubin’s act

On the morning of the call, inside Epstein’s office, one correspondent wrote to another with instructions for dialling Epstein in. Every name in the headers of that sheet is blacked out, and neither correspondent is identified here. The instruction itself contains one forename, belonging to a private individual, which is withheld and marked in the quotation below. It reads:

Sounds like you know what to do but just in case- jeffrey is showing you he has a Conference call today at 12:30pm. The dial-in details are in the text. You really have to read from the bottom up to make sure you k ow all the details. Remind jeffrey of the conference call at 11:30am in an email (example: reminder of 12:30 Conf Call with Jeremy et al. In one hour) then at 12:30 you ask him if he is ready for you to dial in for him. He will say yes. You dial the number and follow the voice instructions. You will punch in the pin code. When it prompts you to say your name say: [a forename] on behalf of Jeff. DO NOT say his last name. All this being said, Jeffrey might dial in himself!!

The reply, at 9:58 a.m., is “Thank you very much! Those details are very important.. / I will know!”16

⚠ This is the office’s own act and its own wording. Rubin is not a party to it, is not copied on it, and nothing on the sheet suggests he saw it. The instruction was written by one Epstein correspondent to another on the morning of a call Rubin had arranged, and the page sets it here because it shows what the office did, not because it shows what Rubin did. Neither the sentence “DO NOT say his last name” nor the decision behind it is his.

What the fund was, on the public record

Ten days before the “Fund Optics” message, the name had been registered in the United States Virgin Islands, to one of Epstein’s own companies. A certificate of the Office of the Lieutenant Governor, Division of Corporations and Trademarks, records:

Trade NameDEPLOY CAPITAL
Nature of BusinessBUSINESS MANAGEMENT & CONSULTING
Registrant(s)SOUTHERN TRUST COMPANY, INC.
Registration DateJuly 6, 2018
Expiration DateJuly 6, 2020

It is witnessed under seal on 13 July 2018 and signed by the Lieutenant Governor.40 Southern Trust Company, Inc. is an Epstein company, and has its own page here.

⚠ What this does and does not establish. The registrant is the company, not this subject. No document read for this page shows him registering the name, being told of the registration, or holding any interest in Southern Trust Company, and the certificate names him nowhere. What it establishes is about the name: on the public record of the territory, “Deploy Capital” was a trade name of a company of Epstein’s, from ten days before the exchange set out below.

The page sets the two dates side by side and draws no inference between them. The registration is 6 July 2018 and the certificate 13 July; the message proposing that Epstein be described to prospective investments as “a strategic LP” is 16 July. Whether the writer of that message knew what the certificate says is not established by anything read here.

August 2018: “it can t be questionable ethics”

The Layer One deal did not close. Through mid-August 2018 Rubin chased paperwork (“I asked him for help a few weeks ago in setting up the papers for Layer 1 and he hasn’t been very responsive”; the sheet gives that person only as “Richard”) and warned that “We may lose the Layer 1 deal otherwise; I think they are anxious.”41

On Thursday 23 August 2018 Rubin asked directly: “What should we do with Layer 1?” and, on the line beneath it, “Are we passing? Should we ask them to do more research?” Epstein’s answer, on a sheet that is clean at the image, is the passage that sits against the July exchange:

we are not . jeremy I am more than happy to fund things but as i am high profile , it can t be questionable ethics .

Rubin replied that “while ethics may be grey, I have no reason to suspect immorality from the team” and that he would “focus in on deals which are a little later stage/with stronger track record/compliance”. Epstein then wrote “their deal is to pump the currency , it is dangerous”. Rubin answered:

There’s a grey area between pump and develop.

In theory blockstream is the same kind of bet; their business model is largely appreciation in Bitcoin price. However, it is less “marketing” and more that they legitimately improve the fundamentals.

I don’t disagree to back out for these reasons, I’ve notified the team that we are not moving forward.

Epstein’s last message in the thread is a question: “It is safe unless it makes money?”42

Epstein wrote “you are right re appearance” on 16 July 2018 and “I am more than happy to fund things but as i am high profile” on 23 August 2018. Both sentences are his, both are in the release, and both dates are printed unambiguously on their sheets. The sequence is recorded here and no inference is drawn from it.

The last documents

Layer One was not the only thing the fund looked at. Three weeks before it was abandoned, on 1 August 2018, Rubin was copied on a Bitmain proposal and answered “Yeah we should look at this. / Bitmain is a great company, very misunderstood. / Context on the valuation?”43 He went on sending Epstein deals after Layer One was dropped.

The last dated document read for this page is of Thursday 8 November 2018, subject “Boltzmann Analytics”: “Blockchain analytics company I’m adv=sing. They’re fundraising now. Pitch deck, tearsheet sample, and term =heet here:” over a link, and “Let me know what you t=ink.” Epstein forwarded it to Richard Kahn at 12:08 PM by that forward’s own header.844 No document carrying a later date was located by the measurement described below. Epstein was arrested on 6 July 2019.

His public statement

DL News reported on 2 February 2026 that Rubin posted on X that day that he was “glad the emails are being released”, that he had had “some professional engagement” with Epstein which was “never exclusive”, and that “I hope the release of the emails brings us closer to justice for those harmed and a better understanding of the nature of corruption in our society”.18 The Marblehead Current reported the same post in the same terms, adding that he said he had not spoken at length on the subject earlier because he had “realized how little I actually knew” and had not wanted to speculate.19

That statement is about the release as a whole. No statement by him addressing any individual document on this page, and in particular none addressing the July 2018 exchange, was located in any source read for this page, and neither report states that its author put those documents to him. Neither outlet’s text was read against the underlying post, which is why both are attributed above to the outlet rather than quoted as the post.

What this page does not claim

  • It does not claim a total paid. Two cheques were read at the image. The memo on the second states a schedule of four. The first and fourth are attested only by office correspondence, which itself disagrees with a cheque about a date, and neither was read here. No sum is added on this page.
  • It does not resolve the date conflict between the office’s “5-1-2016” and the “2nd quarterly payment” cheque dated 4/19/2016.
  • It does not convert or reconcile any clock. Where a header and a quote-back disagree, both are printed and neither is preferred.
  • It does not state what the July 2018 exchange amounts to. It quotes the proposal, the approval and the script, and identifies each speaker.
  • It does not explain the 2017 gap, or assert a cause for it, or assert that documents were withheld.
  • It does not state the size of Deploy Capital. Rubin’s own words in one message set a cap he was proposing to observe, and a figure can be worked back from it; the cap is his, the arithmetic would be this page’s, and neither is adopted here. No corpus count for the phrase “Deploy Capital” is given on this page, because the phrase returns documents predating the fund and a count of them would not be a count of the fund.
  • It does not identify the correspondent who made the 2014 introduction beyond the forename that Ito’s own reply prints, and it does not identify the signatory of either cheque.
  • It does not name or characterise the private individuals who appear in these threads by forename only.
  • It makes no claim about anything outside the documents cited. It does not say that the correspondence described here was his only work in this period, and the record read for this page contains no account of his work that does not involve Epstein.

How the record was measured

The corpus was searched with exact-phrase queries against the release’s full-text index, which covers the EFTA, HOUSE_OVERSIGHT and DOJ-OGR productions together. "Jeremy Rubin" returns 809 page hits across 726 documents; adding the handle form, "jeremy rubin" OR "jeremyrubin", returns 873 page hits across 733 documents.45

Those are returns of an index, not a curated population, and this page does not claim to have read them all. Sixty document identifiers are cited on this page. Twenty-nine of them were rendered and read at the image, at 400 dpi for the EFTA sheets, at 600 dpi where a single character was in issue, and at its native resolution for the single-sheet House Oversight image; the remaining thirty-one were read in the text layer only and are cited for the existence and date of a document rather than for its wording. Each footnote below says which of the two it was, because whether a fault is on the paper or only in the extraction can be settled nowhere else. ⚠ An earlier revision of this page published two defects that the extraction contains and the sheets do not, a full stop for a hyphen inside a date and a capital letter for a numeral in a company name, each under a footnote certifying the sheet had been read at the image. Both were found on review, both are corrected, and both are recorded in the footnote for the sheet concerned rather than removed silently. Real damage and invented damage sat side by side inside one quoted sentence, which is the reason every quotation on this page is now bound to a magnified crop of the sheet it is taken from. The year distribution in the 2017 section above was derived mechanically from the first header date printed on page 1 of each of the 726, and is reported as such: it is a measurement of what the first page of each sheet prints, not of when each message was sent.

The earliest and latest dates given on this page were taken by sorting every header date found across the whole return, not from the page a search happened to surface. Three mechanisms bound this as a floor rather than a census: a document that refers to him by forename alone, or by a description rather than a name, is invisible to a query keyed on the full name; a document in which a producer barred his name is unreachable by any query; and a message of his quoted inside a thread whose extraction is damaged may not be reachable by any string.

Absence tests were written before they were run. No document located by them names him on a flight log, a passenger manifest or a property record; every document pairing his name with the vocabulary of travel or of the island proved to be an office day sheet on which those lines belong to other entries, or an unrelated article; and no document pairs his name with the vocabulary of the offence. The queries themselves are not printed.20

  • jeffrey-epstein
  • joi-ito: on the copy line of the first introduction and of the first of Rubin’s own introductions
  • madars-virza: introduced by Rubin on 21 October 2015
  • bryan-bishop: introduced by Rubin on 12 November 2015; also the LedgerX sequence of March 2016
  • guerman-aliev: the January 2018 Altpoint Capital sequence that followed Rubin’s introduction of Gerald Banks
  • gratitude-america: the foundation that drew both cheques
  • richard-kahn: sent the consulting agreement, chased the cheques, and was copied on the 2018 fund’s account set-up
  • darren-indyke: an officer of the foundation in the relevant years
  • deutsche-bank: the drawee bank on both cheques
  • lesley-groff: on the copy line of the 2014 introduction and the author of the October 2015 Cambridge day sheet
  • karyna-shuliak: recipient of Epstein’s forward of the July 2016 cheque correction
  • mit-media-lab: the institution through which most of these meetings were arranged
  • the-2018-germline-editing-proposal: a separate 2018 sequence in which Rubin’s opinion of a third party is quoted

Notes

Footnotes

  1. EFTA02477799 p.1 of 1, Bates EFTA_R1_01591978. Read at the image at 400 dpi. Header: “From: Jeremy Rubin ▮ / Sent: Sunday, December 27, 2015 1:42 PM / To: Jeffrey Epstein / Subject: Jeremy’s plans for next year:”. The sender’s address and the value in the signature block are barred. The quoted-printable damage, including the literal <=div>, is printed on the paper and is not repaired. ↩ ↩2 ↩3

  2. EFTA01005408 pp.1 to 2 of 3. Both pages read at the image at 400 dpi, and the figures were taken from the image because the text layer misreads them. Header: “From: Jeremy Rubin <▮> / To: “jeffrey E.” [email protected] / Subject: Re: Investing in Layer One / Date: Sat, 14 Jul 2018 16:34:02 +0000”. At the image p.2 prints “$250k”, “500k for 3.5%” and “$1M now for 10%”; the corpus text layer renders the first as “5250k” and the third as “S1M”, and both are extraction artefacts. “Reutrns” and “projects’s” on p.2 are the writer’s own and are not mended. The quoted chain on p.1 attributes Epstein’s messages to 2:51 AM, 8:14 AM and 9:20 AM and Rubin’s to 3:39 AM, 9:53 AM and 12:12 PM on 14 July 2018, against the +0000 Date: header of the covering message; see [^clocks]. The sender domain renders “grnail.com” in the text layer and is an extraction artefact. ↩ ↩2 ↩3 ↩4

  3. EFTA02602084 p.1 of 1, Bates EFTA_R1_01790290. Read at the image at 400 dpi, and the damage quoted above is printed on the paper. Header: “From: Jeremy Rubin <▮> / Sent: Monday, July 16, 2018 6:29 AM / To: jeffrey E. / Subject: Fund Optics”. The sender’s address is barred. The lines =br> and =/div> are markup fragments printed as body text on the sheet, and the faults in “It’=”, “=oing”, “I’m=concerned”, “spo=ked”, “thi=”, “=o”, “you=are”, “conce=n” and “guidance=&” are likewise on the paper. No clean copy of this message is cited as this identifier; the clean rendering is a different sheet and is footnoted separately. ↩ ↩2 ↩3 ↩4

  4. EFTA02584858 p.1 of 1. Read at the image at 400 dpi. Header: “From: ▮ / Sent: Saturday, June 21, 2014 9:46 PM / To: Jeffrey Epstein / Subject: At a session…”. The sender’s name is barred in the From field and the closing name after “Xoxo” is barred in the body. The quoted-printable fault in “p=oject” is on the face of the sheet and is not repaired. The remainder of the message concerns the correspondent’s own plans and is not quoted here. ↩ ↩2 ↩3

  5. EFTA02703853 p.1 of 1, Bates EFTA_R1_02085559. Read at the image at 400 dpi. Header: “From: Joi Ito ▮ / Sent: Wednesday, June 25, 2014 10:56 AM / To: ▮ / Cc: Jeremy Rubin; Jeffrey Epstein; Lesley Groff / Subject: Re: Introduction / Attachments: signature.asc”. The corpus text layer renders the weekday “Wednesda”; the sheet prints “Wednesday”. The quoted original beneath is attributed “On Jun 24, 2014, at 20:07”, against the Date: header of +0000 on the original’s own sheet; the two are on different clocks and no interval is computed. A closing line giving an alternative address for Ito is barred. ↩ ↩2

  6. HOUSE_OVERSIGHT_001953 p.1 of 1. Read at the image, which in this production is a single JPEG per sheet at approximately 300 ppi and is read at its native resolution. “From: Lesley Groff”, her address following the name in square brackets and withheld here, then “Sent: 10/9/2015 10:18:17 PM / To: Jeffrey Epstein [[email protected]] / Subject: Tentative Cambridge Schedule / Importance: High”. The schedule is headed “Fri. Oct. 16, 2015 NY-Cambridge” and every entry on it is marked TENTATIVE. A short value after the 11:00am entry is barred. ↩ ↩2

  7. EFTA00367331 p.1 of 1. Read at the image at 400 dpi. Header: “From: ▮ <▮> / To: Jeremy Rubin <▮>, Jeffrey Epstein <▮> / Cc: Lesley Groff <▮>, Joichi Ito <▮> / Subject: Introduction / Date: Wed, 25 Jun 2014 00:07:41 +0000”. The sender’s display name, the sender’s address, every recipient address and the closing signature are barred. The body is clean on the sheet and is quoted in full above. ↩ ↩2

  8. EFTA02615696 p.1 of 1, Bates EFTA_R1_01815523. Read at the image at 400 dpi. Header: “From: Jeremy Rubin ▮ / Sent: Thursday, November 8, 2018 6:47 AM / To: jeffrey E. / Subject: Boltzmann Analytics”. The quoted-printable damage, including the literal <=iv class="gmail_default"...> opening, is on the face of the sheet. A line beneath the link and the signature block are barred. ↩ ↩2

  9. EFTA02477825 p.1 of 1, Bates EFTA_R1_01592007. Read at the image at 400 dpi. Header: “From: jeffrey E. [email protected] / Sent: Sunday, December 27, 2015 1:44 PM / To: Jeremy Rubin / Subject: Re: Jeremy’s plans for next year:”. The damage in “=tructure” is on the face of the sheet. The quoted original beneath is attributed “On Sun, Dec 27, 2015 at 9:41 A=” and carries its own, differently damaged rendering of Rubin’s message; see [^clocks]. ↩ ↩2

  10. EFTA02363644 pp.1 to 3 of 4, Bates EFTA_R1_01357082 to EFTA_R1_01357084. pp.1 to 3 read at the image at 400 dpi. p.1 header: “From: Richard Kahn ▮ / Sent: Wednesday, March 9, 2016 2:11 AM / To: Jeremy Rubin / Subject: Re: Gratitude America”. The chain beneath runs back through 8 March 2016, 16 February 2016, 4 February 2016 and Epstein’s forward of “Fwd: Info:” to Kahn, “Date: January 24, 2016 a= 8:15:08 PM EST”. Every telephone number, fax number and email address in the signature blocks and headers is barred on the sheet; a line inside the 4 February message naming a typographical form of Rubin’s own email address is present on the sheet and is deliberately not reproduced here. Kahn’s message quoted above carries no Date: line of its own on this sheet; it is the message Rubin answered at 4:52 PM on 4 February 2016. ↩ ↩2 ↩3 ↩4

  11. EFTA01348717 p.1 of 1. Read at the image at 400 dpi; the text layer of this sheet is fragmentary and every value above was taken from the image. Bates DB-SDNY-0035638 and SDNY_GM_00181822; the sheet is stamped “CONFIDENTIAL - PURSUANT TO FED. R. CRIM. P. 6(e)”. Drawer “GRATITUDE AMERICA, LTD”; drawee “DEUTSCHE BANK TRUST COMPANY AMERICAS, NEW YORK, NEW YORK 10154”; cheque number 362; date 4/19/2016; “PAY TO THE ORDER OF Jeremy Rubin”; ”$ **25,000.00”; “Twenty-Five Thousand and 00/100”; memo “2nd quarterly payment”. In the payee address block the street line is blacked out and the town line reads Cambridge, MA; the street line is not reproduced and is not reconstructed. The authorised-signature line carries a handwritten signature; this page does not identify the signatory. A further black bar covers part of the MICR line. ↩ ↩2 ↩3

  12. EFTA01348712 p.1 of 1. Read at the image at 400 dpi; as with cheque 362, every value was taken from the image. Bates DB-SDNY-0035632 and SDNY_GM_00181816, stamped “CONFIDENTIAL - PURSUANT TO FED. R. CRIM. P. 6(e)”. Same drawer and drawee; cheque number 365; date 8/1/2016; “PAY TO THE ORDER OF Jeremy Rubin”; ”$ **25,000.00”; “Twenty-Five Thousand and 00/100”; memo “3rd Quarterly Payment of 4”. In the payee address block the street line is blacked out and the town line reads Marblehead, MA; the street line is not reproduced and is not reconstructed. The authorised-signature line carries a handwritten signature, not identified here. A black bar covers the MICR line. A cheque of the same date and the same memo formula drawn to another payee, EFTA01348714 p.1, read from the text layer only, shows the same “of 4” schedule in use for a second recipient. ↩ ↩2 ↩3

  13. EFTA01791081 pp.1 to 2 of 2, Bates EFTA_R1_00120694 to EFTA_R1_00120695. Both pages read at the image, p.1 at 600 dpi and p.2 at 600 dpi. p.1 header: “From: Richard Kahn ▮ / Sent: Thursday, July 7, 2016 1:34 PM / To: jeffrey E. / Subject: Jeremy rubin / leon botstein”. The quoted-printable damage that is on the paper is “=ubin”, “=elow”, ”=-27-2016”, “=hecking”, ”=-1-2016” and “=oth”, and it is not repaired. ⚠ The second date in that sentence prints “5-1-2016”, with a hyphen, at the image. Both the served corpus and the local extraction render it “5.1-2016”, and an earlier revision of this page published that form as though it were the sheet’s. It is not: it is the extraction’s, and it has been corrected. A large black bar follows Kahn’s signature block. The quoted message of 27 June 2016 is from a named office correspondent whose address is barred; the message quoted on p.2 is dated “June 14, 2016 at 2:21:24 PM =DT” and its sender is given there by forename and an initial, with the address barred. The 14 June sentence is quoted above only in the two spans that carry its content, because the three rules that punctuate it could not be certified at the image as en-rules rather than em-rules and this page does not publish a character it cannot certify. Both the served corpus and the local extraction render all three as em-rules and both carry the full sentence; an earlier revision of this page asserted that the extraction dropped a clause, and that assertion was wrong against both. ↩ ↩2 ↩3

  14. EFTA02602102 p.1 of 1, Bates EFTA_R1_01790328. Read at the image at 400 dpi, and the damage quoted above is printed on the paper. Header: “From: jeffrey E. [email protected] / Sent: Monday, July 16, 2018 10:22 AM / To: Jeremy Rubin / Subject: Re: Fund Optics”. The sheet prints “=houdl” and “v=ncenzo”, and a trailing ”=” after “good call”; neither is repaired here. After “maybe skype.” the sheet prints one glyph the render does not resolve, followed by the characters “=A0”. Neither is reproduced inside the quotation above, because substituting a character for an unresolved glyph would be a transcription this page cannot defend; the omission is recorded here instead. Beneath, the quote-line reads “On Mon= Jul 16, 2018 at 2:29 AM, Jeremy Rubin <▮></=>> wrote:” and the copy of Rubin’s message inside it is damaged in different places from the copy on his own sheet. A trailer of Gmail metadata prints at the foot. ↩ ↩2 ↩3

  15. EFTA01005111 pp.1 and 2 of 3 (page 3 carries only repeated confidentiality footers). Both pages read at the image at 400 dpi. The clean copy of Rubin’s message of 16 July 2018 begins on p.1 and its last three sentences print on p.2, which is where the three sentences quoted above after “(especially given SF politics)” are to be found. Header: “From: Jeremy Rubin <▮> / To: “jeffrey E.” [email protected] / Subject: Re: Fund Optics / Date: Wed, 18 Jul 2018 15:34:34 +0000”. This sheet carries the whole 16 to 18 July chain and prints every message in it clean, including Rubin’s message of 16 July and Epstein’s reply, both of which are damaged on their own separately produced sheets. Quote-line attributions on this sheet: Epstein 18 July 5:24 AM; Rubin 17 July 6:49 PM; Epstein 17 July 2:23 PM; Rubin 17 July 5:15 PM; Epstein 16 July 3:22 AM; Rubin 16 July 2:29 AM. Exactly one of those is out of order against the message it answers: the reply attributed to 17 July 2:23 PM answers a message attributed to 17 July 5:15 PM. The other five run in order. See [^clocks]. “confernece” is the writer’s own and is not mended. Rubin’s sender address is barred at each occurrence, and the two telephone numbers given in the 17 July 6:49 PM message are blacked out on the sheet; the two forenames beside them are on the sheet and are not reproduced here. Further copies of the same chain exist in the release and are not enumerated. ↩ ↩2 ↩3 ↩4 ↩5 ↩6

  16. EFTA02252128 pp.1 and 2 of 5, Bates EFTA_R1_01040378 to EFTA_R1_01040379. Both pages read at the image at 400 dpi. p.1 header: “To: ▮ / From: ▮ / Sent: Thur 7/19/2018 1:59:55 PM / Subject: Re: Fund Optics”. Every sender and recipient name on this sheet is blacked out, including the name in the salutation of the 9:56 a.m. message, and the correspondents are not described here beyond the fact that the exchange runs inside Epstein’s office. The forename that the instruction directs the reader to speak aloud is a private individual’s and is withheld above, as the other forenames on this page are withheld; the bracketed substitution in the quotation marks where it stands. “k ow” and “Jeffrey might dial in himself!!” are the sheet’s own. Two glyphs the render does not resolve print near the top of the sheet and after “himself!!”, and neither is reproduced. p.2 carries the 16 to 19 July chain, and prints “confernece call line pref”. The “Call Info:” block, and the two telephone numbers in the 17 July 6:49 PM message, are blacked out on p.2, and none of those values is reproduced here. ↩ ↩2 ↩3 ↩4

  17. EFTA01006134 p.1 of 4. Read at the image at 400 dpi. Header: “From: “jeffrey E.” [email protected] / To: Jeremy Rubin <▮> / Subject: Re: Fund Optics / Date: Thu, 19 Jul 2018 09:32:49 +0000”. The body is the four words quoted above. This page cites this sheet for that sentence and for nothing else, and quotes nothing else from it. It is the only document in the release that carries Epstein’s answer to Rubin’s question of 19 July; no copy of it was located in which that answer appears, so the sentence could not be taken from another sheet. Nothing else on this page rests on this identifier. ↩ ↩2

  18. DL News, “Epstein files reveal sex offender’s attempts to steer Bitcoin development”, https://www.dlnews.com/articles/people-culture/epstein-files-reveal-desire-to-steer-bitcoin-via-its-developers/, read 2026-09-28. Cited for what it reports, not as a primary source. It attributes the quoted words to a post by Rubin on X on 2 February 2026 and does not state that it put any individual document to him. ↩ ↩2

  19. Marblehead Current, “New Epstein records show payments, crypto talks with former Marblehead resident”, 16 February 2026, https://marbleheadcurrent.org/2026/02/16/new-epstein-records-show-payments-crypto-talks-with-former-marblehead-resident/, read 2026-09-28. Cited for what it reports, not as a primary source. It attributes the same quoted words to the same post of 2 February 2026 and does not state whether it sought comment. It reports three cheques to Rubin, dated February 3, April 19 and August 1, 2016. Two cheques were located and read at the image for this page and a third was not, and the difference between the outlet’s February 3 and the office correspondence’s “2-1-2016” is a further instance of the date conflict recorded in the body. The outlet also reports a registration date and corporate form for Deploy Capital; no primary document establishing either was read for this page and neither is asserted here. ↩ ↩2

  20. Absence tests, each written before it was run, run 2026-09-28. The queries are not printed, and the reason is specific to these queries rather than general. Other pages on this wiki print their absence searches in full, defensibly, because those searches pair a name with financial vocabulary. These pair a living man’s name with the vocabulary of a sex offence. Publishing that pairing is the harm the rule exists to prevent, and it is a harm whether the search returns anything or not; the result is therefore reported and the query withheld. What they settled. No document returned by them places him on a flight log, a passenger manifest or a property record. Fifteen documents pair his name with the vocabulary of travel, of a manifest or of the island, and every one was opened. They fall into four classes and no other: three are copies of an exchange about a mahjong tournament held on an island of that name; two are Epstein-office day sheets on which the travel lines belong to other entries and other people; six are scheduling threads in which the flight referred to is a third party’s, or, on EFTA02249483, Rubin’s own onward travel from New York, mentioned by an Epstein correspondent and not recorded on any log or manifest; and four are messages of July 2018 in which the Virgin Islands and other jurisdictions are canvassed as places to domicile a company, including EFTA01005421 and EFTA01005424. None of the fifteen is a flight log, a passenger manifest or a property record, and none places him at Little St. James. Every identifier named in this footnote was read in the text layer and not at the image, and each is cited here for what class of document it is, not for any wording. No document pairs his name with the vocabulary of the offence. Four documents pair his name with words that also occur in that vocabulary. All four were opened individually and are two copies of a forward of a distributed-ledger whitepaper and two copies of a September 2016 exchange about the tax treatment of physical damage, in which the word appears as a term of art in a sentence about who may bring proceedings. None concerns him or anyone else as a complainant. The four are EFTA00820210, EFTA00820210’s sibling EFTA01784815, EFTA01743596 and EFTA01782119. A reviewing commission put this figure at five; the query re-run for this revision returns four, and the four are named here so the count can be settled against the index rather than against either report. Every document in the return carrying a 2018 or 2019 date reference was checked for a date later than 8 November 2018 and none carries one. ↩ ↩2

  21. Clock disagreements, recorded and not reconciled. The release renders the same message on more than one clock and this page converts none of them. Measured instances relied on above: Rubin’s “Fund Optics” message carries Sent: Monday, July 16, 2018 6:29 AM on EFTA02602084 p.1 and is attributed “at 2:29 AM” inside EFTA02602102 p.1; Epstein’s reply carries Sent: Monday, July 16, 2018 10:22 AM on EFTA02602102 p.1 and is attributed “at 3:22 AM” inside EFTA01005111 p.1. The two differences are not the same length. On EFTA01005111 p.1 one message of Epstein’s, attributed to 2:23 PM on 17 July, answers one of Rubin’s attributed to 5:15 PM the same day, and the other five attributions on that sheet run in order. On EFTA01059401 p.1 all three of Epstein’s messages, attributed to 5:03, 5:12 and 5:18 PM, answer messages of Rubin’s attributed to 9:00, 9:11 and 9:17 PM the same day. Further instances: EFTA02477825 p.1 (1:42 PM header against a 9:41 AM quote-back); EFTA02703853 p.1 (a +0000 header of 00:07:41 on 25 June 2014 against a quote-back of “Jun 24, 2014, at 20:07”); EFTA02235703 p.1 (a 3:37 PM header on the sender’s own sheet against a 4:37 PM quote-line); EFTA02301247 p.1 (a 1:34 PM header against a 9:33 AM quote-line); EFTA02615049 p.1 (a 6:47 AM header against a 7:47 AM quote-line). No interval on this page is taken across a header and a quote-back, and the only intervals stated are between two headers produced in the same form. No explanation of the pattern is offered, because none was established at the documents. ↩ ↩2

  22. EFTA02535241 p.1 of 1, Bates EFTA_R1_01680645. Read at the image at 400 dpi. Header: “From: Jeremy Rubin <▮> / Sent: Thursday, January 25, 2018 3:37 PM / To: Gerald Banks; jeffrey E. / Subject: Intro Gerald <> Jeffrey”. At the image the sheet prints “Hi Jeffrey,” and then, after a blank interval carrying no text and no redaction bar, “Best,” and “Jeremy”. The signature line prints the handle with quoted-printable damage inside the two link values. ↩ ↩2

  23. EFTA02235703 p.1 of 1, Bates EFTA_R1_00996779. Read at the image at 400 dpi. Header: “To: Lesley Groff▮ / From: jeffrey E. / Sent: Thur 1/25/2018 3:42:55 PM / Subject: Fwd: Intro Gerald <> Jeffrey”, enclosing “From: Jeremy Rubin <▮> / Date: Thu, Jan 25, 2018 at 4:37 PM / Subject: Intro Gerald <> Jeffrey / To: Gerald Banks <▮>, jeffrey E. [email protected]”. At the image the body prints “Altpoint Capital” and “learn more”; the corpus text layer renders these as “Ailpoint” and “ream” and those forms are extraction artefacts, not what the sheet prints. The quote-line time of 4:37 PM and the 3:37 PM Sent: header on the sender’s own sheet are on different clocks; see [^clocks]. ↩ ↩2

  24. EFTA02399215 p.1 of 1, read from the text layer and not verified at the image. “Sent: Saturday, June 21, 2014 10:43 PM / To: Jeffrey Epstein / Subject: Also”, body “Joi also sync’ing with Jeremy rubin. Very interesting guy!” The sender’s name is absent from the From field on the sheet. ↩

  25. EFTA01781741 p.1 of 1, Bates EFTA_R1_00101013, read from the text layer for this page and not independently rendered here; the same sheet is cited at the image on Madars Virza’s page, which reproduces the whole message. Header: “From: Jeremy Rubin <▮> / Sent: Wednesday, October 21, 2015 3:04 AM / To: Jeffrey Epstein; Madars Virza / Cc: Joichi Ito / Subject: Skype with Madars:”. The Sent: field carries no zone and none is supplied here. The quoted-printable damage is on the face of the sheet. ↩ ↩2

  26. EFTA02301247 pp.1 to 2 of 2, Bates EFTA_R1_01168988 to EFTA_R1_01168989. p.1 read at the image at 400 dpi; p.2 read from the text layer. Header: “From: jeffrey E. [email protected] / Sent: Thursday, July 7, 2016 1:40 PM / To: Karyna Shuliak / Subject: Fwd: jeremy rubin / leon botstein”. The body above the forward is the single line “see what happens when no one punished”. The forwarded message is attributed “Date= Thu, Jul 7, 2016 at 9:33 AM” against the 1:34 PM Sent: header on the sender’s own sheet; see [^clocks]. On this copy that date prints as “2-1-2016”. Kahn’s signature block on this copy prints his firm name and a city line, with a black bar beneath. ↩ ↩2

  27. The search for the first cheque, run 2026-09-29. Queries against the served corpus, FTS5 phrase form: "1st quarterly payment" returns 0 documents; "Quarterly Payment of 4" returns 3, which are the two copies of the August cheque to this payee and one cheque of the same date to a different payee; "Jeremy Rubin" AND "Gratitude America" returns 6, none of them a February 2016 instrument. The desk’s own tabulation of the bank production’s Gratitude America cheque images lists nothing between 18 March 2016 and 19 April 2016. This is a negative about the release and about these queries, not a claim that no such cheque was written — the office correspondence says two cheques were cashed before the August one, and this page takes that as it stands. The outside report that supplies the 3 February date is cited as the lead that prompted the search and not as authority for the date. ↩

  28. EFTA02365389 p.1 of 1, Bates EFTA_R1_01360669, read from the text layer and not verified at the image for this page; the same sheet is cited on Bryan Bishop’s page. Header: “From: Jeremy Rubin / Sent: Thursday, November 12, 2015 4:58 PM / To: Jeffrey Epstein; Joichi Ito; Bryan Bishop / Subject: CFTC to regulate bitcoin”. The Sent: field carries no zone and no clock time from it is used here. The damage in “som=” is on the face of the sheet. ↩

  29. EFTA01784295 p.1 of 1, Bates EFTA_R1_00105505. Read at the image at 400 dpi. Header: “From: Jeremy Rubin ▮ / Sent: Thursday, June 16, 2016 10:36 PM / To: Jeffrey Epstein; Suhas Daftuar / Subject: Intro:”. The sender’s address is barred. The quoted-printable damage, including the literal <=iv class="gmail_default"...> and <=div> lines, is on the face of the sheet. The sheet prints the surname of the second named founder as “Morcos”. ↩

  30. EFTA02461765 p.1 of 1, Bates EFTA_R1_01567974, read from the text layer and not verified at the image. “From: jeffrey E. [email protected] / Sent: Thursday, June 16, 2016 10:39 PM / To: Jeremy Rubin / Cc: Suhas Daftuar / Subject: Re: Intro:”, body “Thx look forward to it”. Both Outlook Sent: headers are on the same clock and the three-minute interval is taken between them and not across a quote-back. ↩

  31. EFTA02532013 p.1 of 1, Bates EFTA_R1_01675706, read from the text layer and not verified at the image. “From: Jeremy Rubin / Sent: Monday, February 5, 2018 11:49 PM / To: Jeffrey E. / Subject: How was the meeting?”, body “Hey Jeffrey, / Curious for your thought. / Best, / Jeremy”, with quoted-printable damage in the extraction. ↩

  32. EFTA02531087 p.1 of 1, Bates EFTA_R1_01674322. Read at the image at 400 dpi. Header: “From: jeffrey E. [email protected] / Sent: Tuesday, February 6, 2018 12:40 AM / To: Jeremy Rubin / Subject: Re: How was the meeting?”. At the image the three lines “I liked him”, “How was the meeting with Gerald?” and “Curious for your thought.” print flush left and unindented, with no wrote: attribution line between them, so the sheet does not itself mark where Epstein’s text ends and the quoted text begins. The reading offered above goes no further than the sheet: the last of the three matches the body of Rubin’s message of the previous evening. ↩

  33. EFTA00677002 pp.1 to 2 of 4, read from the text layer and not verified at the image for this page. “From: Jeremy Rubin / To: “Jeffrey E.” [email protected] / Subject: Re: Intro: Jeremy Zach at LedgerX / Date: Sat, 26 Mar 2016 15:32:41 +0000”, carrying the whole chain back to Bryan Bishop’s introduction of Rubin to LedgerX’s chief technology officer on 23 March 2016. The subject line as printed in this copy’s header is itself damaged. The same sequence is cited at the image on Bishop’s page from the separately produced copies, and this page relies on that page for it rather than restating it. ↩

  34. EFTA00638891 p.1 of 1. Read at the image at 400 dpi. Header: “From: Jeremy Rubin <▮> / To: Jeffrey Epstein [email protected] / Subject: Housing Idea / Date: Fri, 30 Sep 2016 05:30:05 +0000”. The sheet is clean; the corpus text layer renders the sender address domain as “gmail(i?gmail.com” and that is an extraction artefact. The quotation above is the closing paragraph of the message; the seven numbered steps preceding it are summarised rather than quoted. ↩

  35. EFTA00655693 p.1 of 1. Read at the image at 400 dpi. Header: “From: Jeremy Rubin <▮> / To: “jeffrey E.” [email protected] / Subject: Question / Date: Fri, 01 Sep 2017 21:52:11 +0000”. The sheet is clean; the signature block beneath is barred. A separately produced Outlook copy, EFTA02360865 p.1, read from the text layer only, prints the same message damaged and truncated. ↩

  36. EFTA01036150 p.1 of 1. Read at the image at 400 dpi. Header: “From: “Jeffrey E.” [email protected] / To: Jeremy Rubin <▮> / Subject: Re: Question / Date: Fri, 01 Sep 2017 22:47:07 +0000”. The double comma and the spacing in the quotation are the sheet’s own. A first reply of the same hour, EFTA01036149 p.1, read from the text layer only, reads “ira withdrawals an issue but first what does it mean to exchange for purchase . do you mean buy ?”; a third, EFTA01036155 p.1, text layer only, reads “feel freee to call”. ↩

  37. EFTA01059401 pp.1 to 2 of 3. p.1 read at the image at 400 dpi; p.2 read from the text layer. Header: “From: Jeremy Rubin <▮> / To: “jeffrey E.” [email protected] / Subject: Re: Renewing Contract for 2017 / Date: Fri, 09 Dec 2016 01:34:01 +0000”. The whole chain of 8 December 2016 is quoted back inside it. On that sheet Epstein’s messages are attributed to 5:03, 5:12 and 5:18 PM and Rubin’s to 9:00, 9:11 and 9:17 PM the same day, so each reply is stamped before the message it answers; see [^clocks]. Every sender address but Epstein’s is barred. The sender domain renders “jeevacation®gmail.com” in the text layer and is an extraction artefact. A further copy: EFTA01059874. ↩

  38. Year distribution derived 2026-09-28 from the 726 documents returned by "Jeremy Rubin", by taking the first Sent: or Date: header date printed on page 1 of each. Counts: 2014, 63; 2015, 99; 2016, 342; 2017, 16; 2018, 175; no header date found on page 1, 31. The six values sum to 726. The sixteen documents with a 2017 header date are, in full: the February 2017 “Greetings from Japan” thread, EFTA00690572, EFTA00690602, EFTA01055270, EFTA01055276, EFTA01055510, HOUSE_OVERSIGHT_029087, HOUSE_OVERSIGHT_029090 and HOUSE_OVERSIGHT_029092; and the September 2017 “Question” thread, EFTA00655693, EFTA01036149, EFTA01036150, EFTA01036155, EFTA02360865, EFTA02639401, EFTA02639406 and EFTA02639428. Each of the sixteen was read at page 1 in the text layer and assigned to one of the two threads; none belongs to a third. They are eight documents per thread, comprising the messages of the thread and separately produced copies of them, and this page does not describe them as sixteen copies of two documents. This note records a distribution and nothing about why it takes that shape. ↩

  39. EFTA02561505 p.1 of 2, Bates EFTA_R1_01723114. Read at the image at 400 dpi. Header: “From: Jeremy Rubin ▮ / Sent: Monday, June 25, 2018 6:55 PM / To: jeffrey E. / Cc: Richard Kahn / Subject: Re: Exchange/OTC Accounts Set Up”. The quoted-printable damage is on the face of the sheet. The signature block and the sender’s address are barred. The quoted chain attributes Epstein’s message to “Mon, Jun 25, 2018 at 11:43 A=” and Rubin’s original to “Mon, J=n 25, 2018 at 8:40 PM”; those are quote-back times and are not reconciled with the 6:55 PM Sent: header. ↩

  40. Certificate of Trade Name Registration, Government of the United States Virgin Islands, Office of the Lieutenant Governor, Division of Corporations and Trademarks. EFTA00798995 p.1, the whole run. Rendered and read at the image at 400 dpi by the desk. The table is reproduced above field for field. ⚠ The tracking number is given here as the sheet prints it, CORPTN-579-2018STT; the corpus extraction renders it CORPTN-579.4018STT, which is an extraction fault and is not reproduced. The certificate carries the territorial seal in gold and is signed by the Lieutenant Governor, Osbert E. Potter, over the words “Witness my hand and the seal of the Lieutenant Governor of the United States, Virgin Islands, this 13th day of July, 2018.” One short value in that witness line is covered on the sheet and nothing is supplied for it here. The registrant’s physical and mailing addresses are printed on the certificate as the registered addresses of a company and are not reproduced on this page. This document names no natural person other than the signing officer, and it does not name this page’s subject. An outside report of 16 February 2026 stated the registration date and registrant before the desk located the certificate; the certificate is a primary record and the page rests on it rather than on that report. ↩

  41. EFTA01007949 p.1 of 3. Read at the image at 400 dpi. Header: “From: Jeremy Rubin <▮> / To: “jeffrey E.” [email protected] / Subject: Re: Closing Layer1 / Date: Mon, 20 Aug 2018 20:07:11 +0000”. The chain runs back through 18, 17 and 15 August 2018. The sheet gives the person Rubin says he asked for help only as “Richard”, and no surname is supplied here. Rubin’s address is barred throughout. The sender domain renders “grnail.com” in parts of the text layer and is an extraction artefact. ↩

  42. EFTA01007544 p.1 of 2. Read at the image at 400 dpi and again at 600 dpi, and the sheet is clean. ⚠ The sheet prints “What should we do with Layer 1?” with the numeral 1. Both the served corpus and the local extraction render it “Layer I” with a capital letter, and an earlier revision of this page published that form; it has been corrected, and the line beneath it, which that revision cut in half without marking the cut, is now given whole. Header: “From: “Jeffrey E.” [email protected] / To: Jeremy Rubin <▮> / Subject: Re: What are next steps? / Date: Thu, 23 Aug 2018 23:19:07 +0000”. The chain of 23 August 2018 is quoted back in full, with Epstein attributed at 2:55 PM and 3:46 PM and Rubin at 1:33 PM, 6:33 PM and 7:05 PM. The spacing inside Epstein’s quoted sentences is the sheet’s own. “a.long” on this sheet is the writer’s own and is not mended. ↩

  43. EFTA01004511 p.1 of 1, read from the text layer and not verified at the image. “From: Jeremy Rubin ▮ / To: Vincenzo Iozzo ▮ / Cc: “jeffrey E.” [email protected] / Subject: Re: Bitmain investment (time sensitive) / Date: Wed, 01 Aug 2018 21:36:25 +0000”. Only Rubin’s own three sentences are quoted above. The figure named in the message quoted beneath is not reproduced here, because this sheet was not read at the image and every figure on this page was taken from an image. ↩

  44. EFTA02615049 p.1, read from the text layer and not verified at the image. “From: ▮ [email protected] / Sent: Thursday, November 8, 2018 12:08 PM / To: Rich Kahn / Subject: Fwd: Boltzmann Analytics”, enclosing Rubin’s message of the same morning. The quote-line inside it attributes Rubin’s message to 7:47 AM against the 6:47 AM Sent: header on his own sheet; see [^clocks]. ↩

  45. Measurement run 2026-09-28 against the release’s full-text index, which covers the EFTA, HOUSE_OVERSIGHT and DOJ-OGR productions together. The queries are given in the body: "Jeremy Rubin", returning 809 page hits across 726 distinct documents, and "jeremy rubin" OR "jeremyrubin", returning 873 page hits across 733. Both are counts of index returns with no limit applied, and neither is a claim that every return is a distinct message or that every return was read. Sixty document identifiers are cited on this page and twenty-nine of those were rendered and read at the image; the per-document position is stated in each footnote above. A count for the phrase “Deploy Capital” was taken and is deliberately not published, because the phrase’s returns include documents predating the fund and a count of them would misdescribe what it measured. ↩

Backlinks