What Exactly Happened?
A company's CEO sent an email asking for help with a confidential financial deal. Once the employee signed an NDA, a call came in from the CEO. The voice sounded familiar, and on the video call the employee could even see the CEO's face.
After that, the employee spoke with a lawyer from a big law firm who'd been introduced to them, and even went over the legal process with them.
But it turned out that the CEO and the lawyer the employee had been dealing with were both fake. A criminal group had used AI to impersonate the voices and images of real company executives and big-law-firm lawyers.
✦ The Case at a Glance
Announced in New York on November 5, 2026
This case, announced in New York on November 5, 2026, involves total damage of 'at least' $129 million.
Defendants Jacob Soall, Sharon Hazan and Guy Sheetrit, along with their accomplices, were indicted for crimes against thousands of companies from July 2024 to February 2026.
Two Identities, Both Faked With AI: The CEO and the Lawyer
In this case, the criminal group used AI to take on the identities of CEOs at several companies and of lawyers at big law firms.
First, they posed as a CEO and called an employee who had the authority to move money. On the call, AI mimicked the CEO's voice and, when needed, even imitated the face. The actual indictment says it wasn't just voice calls. Video calls were used too.
✦ What Made the Fake Lawyers Convincing
Real people, real firms, cloned faces and voices
Once the employees started to fall for it, the next step was to connect them with a lawyer. And not just any fake lawyer. They used AI to impersonate a lawyer who really works at an international law firm.
The lawyers they impersonated were real people at real law firms, often firms among the world's 50 largest. The gang mostly used the identities of people working in those firms' European offices.
They used AI to clone the faces and voices of lawyers who actually exist, and added the backdrop of a real, famous law firm on top. No wonder it was hard for the employees to tell what was real.
The part of this case that stands out most to me is the handwritten document that defendant Jacob Soall kept. Its title is 'Lawyer Call.' It was a script for acting like a real lawyer on calls with employees. The contents are pretty meticulous.
✦ NDA: What's That?
A Non-Disclosure Agreement, in plain terms
The script said to mention that the acquisition target is a listed company and to explain that laws and regulations required keeping the deal secret.
It even brought up insider trading, as part of a strategy to stop the employee who fell for it from telling coworkers about the deal.
It also had a prepared answer for when the employee who'd fallen for it asked, 'Did the other employees sign the NDA too?' The suggested reply was 'answer is not yet . . . although that is likely to change in coming days . . . internal memo coming days.'
According to prosecutors, Soall used this script mostly to pose as a British lawyer.
Disclosed Losses Alone Come to About $31.94 Million
Only seven concrete cases have been disclosed in this case: six with actual losses and one where the transfer was blocked.
✦ The Seven Disclosed Cases
In order, from October 2024 to June 2025
The biggest hit was a fintech company in the United Kingdom. About $11.58 million left that one company alone.
Besides that, the criminals hit hospitals, banks, investment firms and even a pet food company. They went after businesses where employees could authorize large transfers. The New York investment bank case was a little different.
The employee got as far as approving a transfer of about $2.85 million, but luckily the company's internal review blocked it. (Somebody up there was looking out for that employee...)
Anyway, these six companies are the ones with confirmed losses, and the crimes reached thousands of other companies too.
✦ Following the Money
Where the funds went, step by step
The money that left the companies moved to accounts in countries including the United States, Spain, Portugal, Greece, the United Arab Emirates, Singapore, Austria, Slovakia and China.
After receiving it in accounts across several countries, the gang converted it into cryptocurrency.
The indictment says the goal was to hide where the money came from and who owned it. After that, the money was reportedly split among the defendants and their accomplices.
All three defendants ended up arrested in Europe.
✦ Who Was Arrested, and Where
All in Europe, between June and September 2026
Jacob Soall was arrested in Spain on June 17, 2026, Sharon Hazan in Spain on August 5, and Guy Sheetrit in the United Kingdom on September 7.
Soall mainly handled the lawyer impersonation, while Hazan and Sheetrit are accused of being involved in laundering the criminal proceeds.
Beyond Voice Phishing: Welcome to AI Phishing
✦ Same Goal, New Tool
Why this is more than old-school voice phishing
If you look only at the goal of tricking someone into sending money, this is similar to old-school voice phishing. People have long tried to extort money by posing as family, friends, company staff or business partners, right?
But this case used AI to build the belief that 'this is the real thing.'
Younger people have often made videos with AI or watched AI-generated content on YouTube and the like. So they might just be able to tell that the person on the other end of the video call is fake.
But older adults and seniors with little AI experience are far easier to fool.
✦ How Your Own Videos Can Be Used Against You
A hypothetical, one step at a time
Say you run a YouTube channel. Both your face and your voice are right there in your YouTube videos. Or you might have videos with your face and voice posted on your social media accounts.
Then someone can take that data, and we're now in an era where a single personal PC is enough to whip up a deepfake video of you.
If that person video-called your parents, acted like you, and asked for money in your exact voice, could your parents tell?
Or what if a criminal called to say they'd kidnapped you, then showed your parents an image of you tied up tight and let them hear you crying (in a voice identical to the real one)? Could your parents tell it was fake?
It'll probably get harder and harder to tell.
So going forward, I suspect that posting content with your own face and voice on the web is going to become very risky. We've entered an era where it's far too easy to impersonate someone else and commit crimes or do weird things.
That's all for this post. See you next time.
Related posts
- Google SynthID Detector Is Free: Was This Image AI-Made?
- AI Agent Disabled Its Monitor and Lied. Is Yours Safe?
- Qwen Keeps Reaching for Alibaba's Servers: Is Local AI Safe?
🌐 Korean version: 사장님들 조심하세요... CEO를 사칭해 1억 2,900만 달러를 도둑질한 AI 사기단의 실체
🌐 Japanese version: シーイーオー偽装で1億2,900万ドルを盗んだAI詐欺団の正体
No comments:
No comments yet. Want to leave the first one?
Post a Comment