Deepfake scams are already beating last year's numbers in a big way. Reported losses in 2026, through Aug. 17, are 263% higher than the total for all of 2025, according to TRM Labs. The firm's AI-in-Crime Adoption Index now marks AI as having reached a 'mature' level of adoption in the scam category — the only crypto-crime category where that's true.
How fast AI fraud is growing
TRM Labs' broader series tracking all scam reports that mention AI has increased about 25-fold since 2022, including cases where victims used AI tools themselves. The scarier stat is the one focused on scammer-side adoption: deepfakes, chatbots, and AI-generated lures have climbed roughly 13-fold over the same period.
Chainalysis paints a similar picture. Inflows to impersonation scams rose more than 1,400% year over year, and operations with visible on-chain links to AI service providers generated 4.5 times more revenue on average than those without those links. The growth isn't linear — it's compounding.
The FBI's scoreboard
The FBI's 2025 Internet Crime Report logged 22,364 complaints with an AI-related descriptor, totaling $893.35 million in losses. Complaints involving cryptocurrency descriptors racked up $11.37 billion in losses. So AI fraud is still a slice of the overall crypto-crime pie, but it's a slice that's expanding quickly.
And 2026 isn't even over yet. TRM's data stops in mid-August, meaning the final number for this year could be a lot uglier.
Why deepfakes work so well
Deepfakes don't break in; they get invited. Attackers use them to obtain cooperation from authorized users rather than stealing access. That flips the security playbook on its head, making post-onboarding identity checks increasingly important.
FinCEN has flagged the warning signs for financial institutions: mismatched identity information, suspicious device or location changes, third-party webcam tools, and resistance to multifactor authentication. When someone on a video call refuses to hold a badge, that's a red flag.
The broader threat landscape
TRM Labs' review of first-half crypto hacks found smart-contract vulnerabilities still common, but the biggest losses were in infrastructure and operational compromises. That's a different bucket from AI scams, but it points to a wider shift: crypto crime is getting more sophisticated across the board.
For exchanges and regulators, the takeaway is straightforward. The attack surface now runs from the smart contract to the human on the other side of the screen — and the human side is now a lot easier to fool.
