Seven post-conviction forensic experts — including four former FBI examiners — and an independent expert retained by Newsweek concluded that evidence the prosecution described as “at the heart” of the case had been falsified and planted, with government involvement. You can analyze the affected charges here →
Jack Smith’s expert was David Loveall II, and his report in this case was undated, proofless, and scientifically invalid. See the point-by-point breakdown →
Verify in 5 min
Before a key device was forensically preserved, the government brought in a FBI expert who changed data on the device. The expert did not sign the chain of custody, and the government concealed the expert’s involvement for more than four years after trial. This was one of the two devices the experts concluded had been falsified and planted with files.
Verify in 2 min
The government said the original FBI examiner of that same device was unavailable because he had been sent to Ghana. Ghana immigration records say he was not there.
In their March 26, 2018 detention letter to Judge Garaufis, prosecutors framed the arrest as a Mexican immigration matter and represented that Raniere was a fugitive who had “purposely concealed his location”, requiring six weeks of coordination with Mexican authorities to locate.
EDNY detention letter, Dkt. 8-1 (March 26, 2018) (PDF)
The same EDNY had handed Mexican authorities a file listing Raniere’s home address, workplace, partner, the restaurant he frequented, and a note that he was “known to walk several miles each day at odd times” — with street photographs.
U.S.-provided Mexican immigration file, Dkt. 1273-9 (PDF)
You cannot be in hiding and be that findable.
Receipt is in the FBI’s name. Traveler: KEITH A RANIERE / DOJE2 FBI. Route: PVR→DFW, departing 7:06 a.m. on March 26, 2018 — booked the day before.
CWT travel receipt — FBI purchase of commercial flight (PDF)
A routine Mexican deportation does not involve the FBI booking the commercial seat.
EDNY framed the arrest as Mexican immigration officials encountering an “uncooperative” subject who had “purposely concealed his location” — covering up the EDNY orchestration that actually drove it.
The resulting media coverage portrayed a dramatic international apprehension of a fugitive sex-cult leader — exactly the narrative EDNY needed to seed the rest of the case.
The same EDNY that handed Mexico his walking route told a federal judge he had “purposely concealed his location” — and the FBI bought his flight home before any Mexican order existed.
The same orchestration also broke the court’s sealing order. See: Breaking the Seal →
Search-scene photographs are supposed to document evidence exactly as it was found. That's why investigators first photograph items "in place" — before touching or moving them. Otherwise, there is no way to prove where an object actually came from.
Without that safeguard, investigators could rearrange scenes or introduce new objects after the fact and present the result as authentic.
Click any photograph to enlarge.
On March 27, 2018, FBI agents searched a townhouse at 8 Hale Drive, Halfmoon, NY. The search sought nude photos of adult women allegedly used as blackmail material.
Cameras were important.
They documented two cameras at the scene. They only took one.
The first camera — Item 1 — was photographed in place. Then moved for a clearer ID photo. Collected as evidence.
The second camera — Item 5:
What the jury saw: this camera with its cord draped over it on a cluttered countertop, as if just found there:
But it wasn't actually there:
No "in place" photograph. No record of where it came from.
And in the evidence log, written as "NOT TAKEN."
The jury saw the evidence photos. Looked like they found cameras, as expected.
And the camera they did collect — the one handled by protocol? That's the one whose memory card was later altered in FBI custody. Its files have since been proven falsified — by seven experts (four former FBI) and an independent expert retained by Newsweek.
The warrant sought "sex trafficking" paraphernalia.
Agents collected a book on The History of Torture as Item 3.
Later they placed two books about the central alleged crime — sex trafficking — on the desk.
No item label. No photograph recording where they were found.
The books weren't on that desk:
The photo log says they came from the bookshelf — and the prior photo was of a bookshelf. But in that bookshelf shot, you can't see these books:
The jury saw the search photos. A defendant accused of sex trafficking — with two books on sex trafficking sitting at the townhouse.
If those books were authentically at the property, that would be damning evidence. So why wouldn't agents photograph them in place? Why wouldn't they take them as evidence?
There are three search photos of devices on the bookshelf above the desk in the upstairs office. Two of them are demonstrably an agent-created vignette:
Agent-created vignette
Agent-created vignette
All three photos show the same exact spot on the same bookshelf.
You can confirm this by comparing the wall brackets, the monitor edge, the shelf line, and the wires:
In the Item 2 shot — just three devices on the shelf. In the Item 37 shot — the same shelf, but now: a Rubik's Cube (unopened), a Stem Cell DVD, two CDs, and two books on sex trafficking.
5 objects used to create the vignette. None photographed in place. None collected as evidence.
And the two books? The exact same two books that "appeared" earlier on the desk.
Earlier in the search, the bookshelf held Item 2 in this spot. Then it changed — first to Item 36, then to Item 37.
Here's the same sequence brought to life — with the intermediate steps in between, so you can see the multistep construction of the vignette:
Here's what had to have happened, step by step, between these shots — based on the FBI's own evidence log actions:
By steps 7 and 8, Mills, Mergen, and Hochron all knew they were using a manufactured vignette as if documenting an authentic scene. All three are implicated in scene creation that became search-scene photographs shown to the jury at trial as authentic.
And once you've established this with one of the scenes — how can you be sure any of the scenes the jury saw were authentic? That anything was truly found where the agents said it was?
The arrest complaint was placed under seal by the court. The unsealing order — signed by Hon. Steven M. Gold — is dated March 26, 2018. Until then, the charges were not public.
Lauren Salzman testified at trial that Mexican authorities presented Raniere with the EDNY charges on March 25, 2018 — one day before the court unsealed them.
Mexico had the sealed charges before the court unsealed them. EDNY broke their own court’s sealing order to stage an international takedown.
Item 1B16 — a Western Digital external hard drive (serial WCAS81365334), seized on March 27, 2018 at 8 Hale Drive (Halfmoon, NY). This is the same hard drive on which the alleged child-pornography photos were later "discovered" — the device at the heart of the planting case (see Hook 1).
The FBI's chain-of-custody form is the official, signed record of every person who has handled the evidence: who took it, when, who they returned it to, and when. It is the only document a court has to verify that the device admitted at trial is the same device that was seized — unaltered, untampered, untouched outside the documented transfers. Gaps in the form mean gaps in that verification.
The official chain-of-custody form for Item 1B16 shows SA Michael Lever signing the device out — and then no further entries. No signature recording the device being returned to evidence storage. No subsequent transfers. The form runs blank from Lever's checkout all the way to the bottom of the page.
Hard drive chain of custody, last page — "No Entries Post-February 2019 in the Hard Drive's Chain of Custody."
Full hard-drive chain of custody (PDF)
February 22, 2019 was months before trial began on May 7, 2019. And the chain-of-custody record still shows no entry for its return.
On the verge of trial, a hard drive that had been in FBI custody for eleven months suddenly produced alleged underage nude photos — outside the warrant's timeframe and charges, justified to the court as "plain view because accidental," later admitted on camera by the lead prosecutor to have been targeted, and forensically determined to have been planted.
The hard drive was seized at 8 Hale Drive (Halfmoon, NY). The warrant (Attachment B) was scoped to sex trafficking, forced labor, extortion, and racketeering. Child pornography is not on the list.
8 Hale search warrant, Attachment B (PDF)
On February 21, 2019, the FBI claimed to have accidentally discovered nude photos of a female (no sex act depicted) on the drive, dated to 2005. The 2005 date is the entire basis for charging the images as illegal — it would have made the subject 15. The discovery was logged "pursuant to 8 Hale Warrant" and justified to the court as "plain view." It produced new child-exploitation charges, possession charges, and all five codefendants pleading guilty within a week, on the eve of trial.
Dkt. 618 — "accidental discovery" excerpt (PDF)
"Once we knew about the picture taking, picture keeping, it was a matter of finding those. And when I mean, that is game-changing evidence."
— AUSA Moira Kim Penza, The Vow, S2, Ep. 6, 15:20–15:38 (HBO 2022).
The language of a targeted search — not an accident.
Seven post-conviction forensic experts (four former FBI), with a combined 150+ years of digital forensics experience, concluded under penalty of perjury that the photos had been planted on the drive and that timestamps on both the drive and the Canon memory card had been falsified to simulate the 2005 timeframe.
"The involvement of government personnel in this evidentiary fraud is inescapable — an unprecedented finding in our combined 150+ years of forensic experience."
Joint Forensic Report, Doc. 1253-1 (PDF)
In FBI digital forensics, the original device is supposed to be copied once through a process called "forensic imaging," which creates a complete replica of the data. After that, investigators work from the forensic copy (also called a "forensic image") — not the original device itself. That safeguard exists because repeatedly accessing the original device risks altering or contaminating the data itself.
On June 11, 2019 — in the fifth week of a six-week trial — a secret, prohibited second forensic image of the camera's memory card was created and used by a substitute FBI examiner, Brian Booth. The report from this second image contained 37 additional files that did not appear in the report from the first authorized image, made months earlier — after the card had already been altered once in FBI custody (see Secret Alteration by Technician). Seven post-conviction forensic experts (four former FBI) concluded at least 28 of those 37 files were demonstrably manipulated and at least 20 were planted. The episode implicates SFE Brian Booth and SSA Trenton Schmatz in knowing violations of FBI policy.
Booth substituted in for CART analyst Stephen Flatley on the basis of one claim: that Flatley was unavailable because he had been sent to Ghana. Lead prosecutor Penza told the court Flatley was "in Africa right now"; Booth testified Flatley was "out on assignment in Ghana". Ghana's own immigration records show that wasn't true. Full Ghana drawer →
Booth created a second forensic image of the memory card directly from the original device. Under FBI DEPG 3.3.11.2, a forensic re-examination requires written approval from the Assistant Director of the FBI's Operational Technology Division. That approval was never obtained.
Booth's own CART notes describe his "processing" as "SSA Trenton Schmatz concurred and authorized to process the item" — but a CART supervisor cannot authorize what only an Assistant Director can.
DX 961, Booth CART notes (PDF)
Four former FBI forensic examiners (combined 55 years of FBI experience) reviewed the misconduct and named both Booth and Schmatz as responsible for the unauthorized second image. Their conclusion: "this was a knowing violation of protocol by SFE Booth" and Schmatz's approval "was also a knowing violation of FBI protocol."
Joint Process Report by four former FBI experts, Doc. 1273-8 (PDF)
At trial, AUSA Hajjar framed Booth's work as creating just "another report", with no mention of a second forensic image:
And in a sidebar with the judge, Hajjar represented that Booth "never said I was the one who imaged the devices" — concealing the prohibited second forensic imaging from the court.
On January 27, 2025, AUSA Hajjar acknowledged for the first time that "The Canon Camera was forensically imaged a second time." Her citation? (T:4986–87) — the very transcript excerpt where Hajjar herself had concealed it.
Appeal government opposition, Dkt. Entry 61.1 (PDF)
And to this day, the prosecution still refuses to disclose the actual second forensic image to the defense.
Two forensic copies of the same memory card should contain exactly the same files. They didn't. Booth's report contained 37 additional photo files not in Flatley's first authorized image:
Seven post-conviction forensic experts (four former FBI) concluded at least 28 of those 37 files were intentionally manipulated and at least 20 were planted onto the memory card.
Seven Experts' Joint Forensic Report (PDF) · Flatley index (PDF) · Booth index (PDF)
FBI Senior Computer Scientist out of Quantico. Recipient of the Presidential Early Career Award for Scientists and Engineers for his contributions to digital forensics, the FBI Director's Award for Outstanding Technical Advancement, and the Intelligence Community Seal Medallion.
He is not a forensic examiner routinely assigned to prosecutions. Over a four-year span, the only other case in which Loveall was brought in for testimonial evidence was Jack Smith's Mar-a-Lago prosecution of President Trump.
Jack Smith expert notice, Doc. 257-6 (PDF)
On July 21, 2023, the DOJ brought Loveall in to respond to the findings of seven post-conviction experts (four former FBI examiners) that the camera's memory card and an external hard drive had been falsified. Loveall produced a report denying the findings.
Under Federal Rule of Evidence 702, expert testimony must be based on sufficient facts or data. Loveall's rebuttals contain technical language but no proof — only hypotheticals presented as findings. His own diagrams are illustrations he created, not proof.
A forensic report without underlying proof is not admissible expert opinion — it is assertion.
The seven experts found that 37 files appeared on a second, undisclosed forensic copy of the memory card — indicating either planting in FBI custody between the first and second copy, or omission from the first copy's report. They concluded at least 28 of those files had been tampered with and at least 20 had been planted.
Loveall's full rebuttal:
Loveall report, Doc. 1213-3, pp. 4–5.
What's missing:
The government's own examiner, Brian Booth, testified that the FBI uses hash verification on every case: "Message Digest 5, called the MD5 verification" — if "anything gets changed on a hard drive… this whole function… would be different and we know something has been changed."
Not on the cover. Not in the body. Not anywhere. 28 U.S.C. § 1746 requires a date to legally bind a declarant to the penalty of perjury. Without a date, the declaration has no legal force — the expert is not bound by his own findings.
Loveall report, Doc. 1213-3 (PDF)
Loveall's only claimed independent testing was on "a Dell Dimension 8300-20090330" — a computer he says he procured to test the opposing expert's claim. No such Dell model exists. The "20090330" suffix is a date code (March 30, 2009) lifted from a folder name in the case evidence — not part of any real Dell hardware identifier. Either the testing never happened, or the report's most basic factual claim about its own methodology cannot be verified.
Did FBI Senior Computer Scientist David Loveall — recipient of the Presidential Early Career Award for Scientists and Engineers, the FBI Director's Award for Outstanding Technical Advancement, and the same expert Jack Smith selected to explain hash values to a jury in the prosecution of President Trump — not know that his report contained no proof, no date, and a claim of testing on a Dell model that does not exist?
