Three men charged in US with using AI to impersonate CEOs and lawyers in alleged $129 million fraud
US prosecutors in the Southern District of New York have unsealed an indictment charging Jacob Soall, Sharon Hazan and Guy Sheetrit with using artificial intelligence — called 'super intelligence' in the release — to impersonate the voices and images of corporate executives and major-law-firm attorneys, tricking employees into wiring money to accounts the group controlled. Prosecutors allege the scheme took more than $129 million from companies worldwide between 2024 and 2026. All three were arrested overseas and extradition requests are pending; the charges are allegations and the men are presumed innocent unless proven guilty.
Checking in progress
This story is still being checked. Claims below are published as claims — never dressed as facts — and each one moves only when the evidence earns it.
- Collecting reportsDone
- Grouping the versionsDone
- 3Checking claims2 of 4 claims checked
- 4Human gateStill to come
2 of 4 claims checked so far · step 3 of 4: Checking claims
Key points
- An unsealed SDNY indictment charges Jacob Soall, Sharon Hazan and Guy Sheetrit over an alleged AI impersonation scheme; all three were arrested overseas, with extradition requests pending.
- Prosecutors allege the group cloned CEOs' and lawyers' voices and images to trick employees into wiring money, taking more than $129 million between 2024 and 2026.
- The alleged victims include hospital systems, a bank, a mutual fund company, a pet food company and a UK fintech — none named in the release.
- Each man faces seven counts including wire fraud and money laundering; the charges are allegations and the men are presumed innocent unless proven guilty.
What happened
The case is a large-scale version of the 'CEO fraud' pattern — a fake executive instructs an employee to make an urgent, confidential payment — upgraded, prosecutors allege, with AI-generated voice and image so the 'executive' could appear on calls, and a second layer of fake lawyers to make the transaction look real. The indictment says the group typically began by targeting an employee while pretending to be that company's own CEO.
The US Attorney's release uses the term 'super intelligence' (SI) for the technology, and the FBI's New York field office led the investigation with Spanish and British authorities making the arrests. Prosecutors publicised the case with an explicit warning to businesses to protect themselves against AI-enabled impersonation.
Claim by claim
Why this label: The existence of the indictment, the arrests and the pending extraditions are verified from the prosecutor's own release — the primary source the queue required — with the Guardian's report in agreement.
On the ladder: a claim earns Verified from primary evidence, or from two independent sources that do not copy each other.
Why this label: This is the prosecution's allegation, labelled Not yet verified: the release itself states the indictment's contents are only allegations and the defendants are presumed innocent unless and until proven guilty. No defence statement was in the sources read.
On the ladder: the evidence so far sits low — early, single or social sources. We publish the claim as a claim, never as fact.
Why this label: Each figure is the indictment's figure, carried with its exact amount from the release and labelled as alleged; the victim companies are not named in the release and are not named here.
On the ladder: the evidence so far sits low — early, single or social sources. We publish the claim as a claim, never as fact.
Show all claims (1 more)
Why this label: The charges, the statutory maxima and the prosecutor's own words are verified from the release. Maximum sentences are Congress-prescribed figures the release gives for information; any actual sentence would be set by a judge after a conviction that has not happened.
On the ladder: a claim earns Verified from primary evidence, or from two independent sources that do not copy each other.
What we know / what we don't know yet
What we know
- The indictment was unsealed in the Southern District of New York and the three men were arrested in Spain and the UK, with extradition requests pending (US Attorney's release, read in full).
- Each defendant faces the seven counts listed in the release, with the maxima stated there.
- The release itself states the charges are merely accusations and the defendants are presumed innocent unless and until proven guilty.
What we don't know yet
- Any defence statement, plea or response from Soall, Hazan or Sheetrit — none was in the sources read.
- Whether and when extradition from Spain and the UK will be granted, and any trial date.
- The identities of the victim companies, which the release does not name, and how much of the $129 million has been recovered.
Why it matters
If the indictment's account is proven, a cloned voice and face were enough to move eight-figure sums out of hospitals, a bank and other companies — the control that failed was not a password but a person trusting what they saw and heard. The case will be watched as an early test of how courts and companies respond to AI impersonation at scale, whatever its outcome for these three defendants.
Background
Show the background timeline (4)
The period over which the indictment alleges the impersonation scheme operated.
Hazan was previously convicted in the Southern District of New York of conspiracy to commit bank fraud, per the release.
Soall arrested in Spain (17 June), Hazan in Spain (5 August), Sheetrit in the United Kingdom (7 September).
The indictment is unsealed and the charges announced by the US Attorney for the Southern District of New York.
Update history
No silent edits. Every change is written here, with its time.
- Latest
Published in Checking state as a standard from queue MANUAL 2013 (Guardian, 21:20Z) — worked from the DOJ press release itself, as the queue required. CONTROVERSIAL FLAG for Adisa's visibility: this is an accusation story in live proceedings — published under the standing direction on verified evidence with honest labels, not held for approval. JUDGE AUDIT: (1) Headline, slug, SEO title and summary all state the same outcome — three men CHARGED over an ALLEGED $129 million fraud; 'charged' and 'alleged' survive in all four. (2) Procedural facts (indictment, arrests, charges, maxima) are Verified from the prosecutor's release; the scheme and the loss figures are labelled Not yet verified as the indictment's allegations, and no claim states the men committed the fraud. (3) Every dollar figure is the indictment's exact figure from the release. (5) The defence position is absent from the sources and is declared, not filled in. (6) Photo is a labelled file photo of the Justice Department headquarters, probed 200 at both sizes; video check made — no official DOJ/FBI video of this case was found in this run's searches, none attached. (7) Headline, summary and take labels written in all 13 non-English languages in this run; new story, no French/Arabic full text to rebuild.