From: Sent: Monday, October 19, 2020 10:37 PM To: (USANYS) (USANYS) Cc: (USANYS) Subject: RE: GM -- search warrant returns I'm struggling to think of a way to limit '=' or "jeevaction" more because: • is last name. Because she was Epstein's primary secretary who handled the travel logistics for Epstein, his employees, and the victims who traveled. So it makes sense that her last name has so many hits. Those will include essentially all of the day-to-day communications regarding Epstein's schedule and travel. • The term "jeevacation" is the phrase Epstein used for the majority of his email addresses (e.g., [email protected] [email protected] etc.). So it's unsurprising that this phrase is the highest hit. I suspect that number reflects the majority of Epstein's emails. We ran that search term because we do not know all of the email addresses Epstein used, and we suspected that there were more domains with "jeevacation" email accounts than we knew about. We've already run search terms to cut out spam, though we can certainly try to find more. From: (USANYS) Sent: Monday, October 19, 2020 9:21 PM To: ) Cc: Subject: RE: GM -- search warrant returns (USANYS) (USANYS) C Thanks, this is a helpful start. I would recommend we think about how we can perhaps narrow or focus our search terms a bit, perhaps through combinations/connectors or other means, particularly for common names/words that are likely to generate a significant number of hits. For example, "David" turns up nearly 40,000 hits, and "flight" or "flights" turns up over 100,000 just to pick a few. Also, curious as to why' " turns up 450,000 hits and "jeevacation" (which I'm not familiar with but assume you all are?) turns up 750,000. Those two alone seem to account for the overwhelming majority of documents we've identified, so if there's a way to focus those two in particular, that might be a big help. One thing you might want to do is just review a couple hundred random hits, if you haven't already, to get a bit of a flavor for what we're turning up, as that may help inform our efforts to focus the search. I think we may also have the technology to weed out spam and potentially duplicates, if we haven't already, but defer to others with more experience in those areas.... From: Sent: Monday, October 19, 2020 8:51 PM To: (USANYS) (USANYS) Cc: (USANYS) Subject: RE: GM -- search warrant returns The STR report is attached, in case that's useful. From: Sent: Monday, October 19, 2020 8:44 PM EFTA00089653

To: (USANYS) Cc: (USANYS) (USANYS) Subject: RE: GM -- search warrant returns Thanks, M. Taking these one by one: • We designated certain email accounts responsive, per our discussion, at that yielded 90,000 emails. The balance of the 1.4 million emails/documents comes from search terms. • The database contains all kinds of documents—word, PDF, and excel. I haven't seen any texts, and it's mainly the kinds of things you'd expect to be saved on someone's computer. • I don't have numbers by search term, but will check on that and circle back. We sent a list of terms to and asked him to execute, so I'm not sure about the breakdown. • Yes, Maxwell is a search term and her email account has been designated responsive. From: (USANYS) Sent: Monday, October 19, 2020 8:37 PM To: (USANYS) Cc: (USANYS) Subject: RE: GM -- search warrant returns Thanks, Couple of follow up questions on this, and happy to set up a call if that's easier. First, am I understanding correctly that we ran all of our search terms and came up with 90,000 responsive emails and 1.31 million other responsive documents? Or is the 90,000 figure just email account? Second, and related, with respect to the non-email documents, whatever their volume, do you have a sense of what kinds of documents we're talking about? Word documents, texts, excel spreadsheets, some combination thereof? Third, do you know how many documents hit on each term? If we came up with 1 million hits on the names of the victims, that's one thing, but if we came up with a million hits on the word "flight" or "passenger" we might want to think about whether to narrow the terms a bit. Also, I presume we ran Maxwell's name as one of the co-conspirators and ran any email account of hers we identified as one of the email accounts? As noted, happy to jump on a call tomorrow if that's easier. Thanks, From: Sent: Monday, October 19, 2020 8:14 PM To: (USANYS) (USANYS) (USANYS) Cc: Subject: GM — search warrant returns We've run our responsiveness search terms for the search warrant returns, and they yield a total of 1.4 million documents. As a breakdown, 90,000 of those items are emails we designated responsive by account (for example, . The remaining documents are hits on our search terms, which included the first and last names of every victim and suspected co-conspirators, as well as specific keywords (e.g., passenger, flight). The total database is 1.6 million documents. EFTA00089654

1.4 million documents is obviously a substantial portion of the 1.6 million document database. Given our prior discussions, our understanding is that the office is okay with us designating all of these documents responsive, but please let us know if you'd like us to take a different approach. Thanks, Assistant United States Attorney Southern District of New York One Saint Andrew's Plaza New York, NY 10007 EFTA00089655