You’d think the FBI’s crypto wallets would be the most secure vaults in the digital asset world. You’d be wrong. Very wrong.
Prosecutors this week charged a former FBI counterintelligence supervisor with stealing nearly $1 million in cryptocurrency from wallets that were directly tied to the bureau’s own investigations. Then, in a move that feels ripped from a bad tech thriller, he allegedly turned to ChatGPT for advice on how to invest the loot and relocate to Europe.
This isn’t a Hollywood script. It’s a real case that raises uncomfortable questions about trust, custody, and just how dumb smart criminals can be.
The Inside Job: How a Counterintelligence Agent Turned on His Own Investigations
According to the criminal complaint unsealed in federal court, the defendant — a former FBI agent who supervised counterintelligence operations — used his access to law enforcement systems to identify cryptocurrency wallets that had been seized or controlled during active investigations. Over several months, he allegedly drained those wallets of digital assets, funnelled the funds through a series of mixer services and personal wallets, and attempted to cover his tracks.
The total haul? Approximately $975,000, prosecutors say. A tidy sum, but not life-changing — unless you’re planning to flee the country.
What makes this case stand out isn’t just the brazen insider theft. It’s what happened next. The agent allegedly used OpenAI’s ChatGPT to ask questions like: “How should I invest $1 million in cryptocurrency to avoid detection?” and “What are the best ways to relocate to Europe without attracting attention?”
Yes, really. The same chatbot that helps students write essays and programmers debug code was apparently consulted for a getaway plan. The conversations were logged and later recovered by investigators — a plot twist that would be comedic if it weren’t so incriminating.
ChatGPT as a Getaway Planner: The New Frontier of Crypto Crime
This case is a stark reminder that AI isn’t just a tool for good — it’s also a weapon for the clueless. The defendant’s reliance on ChatGPT suggests either breathtaking overconfidence or a shocking lack of operational security. In either case, it’s a goldmine for prosecutors.
The complaint details how the AI gave generic advice — diversify into stablecoins, use decentralized exchanges, avoid large withdrawals — but also flagged that “relocating to another country with large sums of cryptocurrency could be subject to customs and anti-money laundering checks.” The chatbot, in other words, did a better job of warning him than his own training did.
This isn’t the first time we’ve seen criminals turn to AI for help. But it’s one of the clearest examples of how the technology is being weaponized by amateurs — and how it can backfire spectacularly. The same logs that the defendant thought were private became the smoking gun.
What This Means for Crypto Custody and Trust in Law Enforcement
For the crypto industry, the implications are chilling. If the FBI’s own wallets aren’t safe from an insider — a supervisor, no less — then what hope do ordinary investors have? The incident echoes the Coldcard hack that neared $114 million, where self-custody itself came under siege, but here the threat was from inside the vault.
It also raises questions about how law enforcement manages seized assets. The U.S. government has been steadily building its crypto reserves through forfeiture, and this case suggests the controls are far from watertight. A single bad actor with the right credentials can walk away with nearly a million dollars.
Meanwhile, the broader crypto market has been on edge. Bitcoin recently dropped below $63,000 amid geopolitical tensions and Coldcard losses, and this kind of insider theft story does nothing to restore confidence. It’s a reminder that the biggest risks in crypto aren’t always smart contracts or flash loans — sometimes they’re the people in charge of the keys.
The Broader Implications: AI, Crypto, and the Future of Financial Crime
This case is a microcosm of two converging trends: the rise of AI-assisted crime and the persistent vulnerability of crypto custody. The defendant’s use of ChatGPT for investment and relocation advice is almost laughable, but it’s also a harbinger. As AI tools become more powerful, we’ll see more criminals — sophisticated and otherwise — using them to plan, execute, and cover up their schemes.
Regulators are already scrambling. The SEC and CFTC are eyeing AI’s role in financial fraud, and the DOJ is likely to push for harsher penalties for anyone who uses AI to commit a crime. But the cat is out of the bag. The same large language models that help you write an email can help you launder money — if you’re dumb enough to leave a trail.
As for the defendant, he faces up to 20 years in prison for wire fraud and money laundering. The irony? He might have gotten away with it if he’d just kept his mouth shut — and his search history clean.
For the rest of us, the takeaway is simple: trust no one, not even the FBI. And if you’re going to use ChatGPT for criminal advice, at least use a burner account. But I’m not recommending that. Obviously.
Frequently Asked Questions
What exactly did the former FBI agent steal?
He allegedly stole nearly $1 million in cryptocurrency from wallets that were controlled by the FBI as part of its investigations. The funds included Bitcoin, Ethereum, and other altcoins, according to the criminal complaint.
How did he get caught?
Investigators recovered chat logs from his ChatGPT account, where he had asked for advice on investing the stolen funds and relocating to Europe. The logs provided a direct link between the theft and his planning.
What does this mean for crypto investors?
It’s a stark reminder that even the most secure-seeming custodians can be compromised by insiders. Investors should consider using multi-signature wallets, hardware wallets, and decentralized custody solutions to reduce reliance on any single point of failure.
