CryptoMist Logo
Login
FeaturedMarch 12, 2026

Crypto ATM Losses Hit $333M as AI-Powered Scams Surge

Crypto ATM fraud hit $333 million in 2025 as FBI complaints surged 33%, with AI-powered deepfake scams targeting seniors, per CertiK's latest report.

Crypto ATM Losses Hit $333M as AI-Powered Scams Surge

What to Know

  • $333 million in crypto ATM fraud losses were recorded in the US in 2025, per CertiK
  • The FBI received over 12,000 complaints between January and November 2025 — a 33% jump from the year prior
  • 86% of victims are over 60 years old, though younger adults increasingly appear in romance and investment scams
  • AI-enabled social engineering was 4.5x more profitable than traditional scam methods in 2025

Crypto ATM fraud just had its worst year on record in the United States — $333 million drained from victims in 2025, FBI complaint volumes up 33%, and organized crime running operations that look less like opportunistic theft and more like Fortune 500 logistics. The people building those 45,000 kiosks handed scammers a near-perfect tool, and nobody seems to want to own that.

The Numbers Behind a $333 Million Problem

The FBI logged more than 12,000 complaints tied to crypto ATM fraud between January and November 2025 — a 33% increase over the same period the prior year. Total US losses reached $333 million, according to a new report from cybersecurity firm CertiK, shared with reporters on Thursday. The US hosts 78% of the world's roughly 45,000 cryptocurrency kiosks, making it far and away the primary theater for this category of financial crime.

The machines' design is part of the problem. Cash to crypto in under five minutes, minimal identity checks, and a blockchain ledger that only records the operator-to-destination transfer — not the victim's name. CertiK calls this the 'attribution gap,' and it makes forensic tracing nearly impossible without a court order compelling operators to hand over records. That's a feature scammers have clearly studied.

Crypto ATMs are the lowest-friction extraction channel available to scammers.

— CertiK, Web3 Security Report 2025

Who Gets Hit — and Why AI Is Making It Worse

Around 86% of losses involve victims over 60. Older adults bring what CertiK describes as 'liquid savings,' lower crypto literacy, and social isolation — a combination that makes them disproportionately easy marks. But the demographic picture is shifting: younger adults are showing up more often in romance scams and investment schemes, specifically the long-con 'pig butchering' playbook where fraudsters build fake relationships before directing victims to fake platforms.

The five main approaches CertiK documented: pig butchering, government impersonation, tech support fraud, grandparent scams, and fake fraud recovery offers. What unites all of them is that they work through persuasion, not code. Unlike phishing attacks that compromise private keys or wallet-draining exploits, crypto ATM fraud 'relies entirely on social engineering to induce the victim to perform a voluntary physical action at a kiosk,' the firm stated. You walk up, you insert cash, you send it to a stranger's wallet. No hack required.

AI has turned up the heat considerably. Real-time deepfake synthetic media — live video calls that make a scammer look like a government official or a grandchild in distress — are now embedded in active fraud operations. AI-enabled social engineering scams were 4.5 times more profitable than traditional methods last year. The scam pipeline has industrialized too: CertiK notes that the profile of crypto ATM fraudsters has shifted from lone actors to structured transnational organizations running corporate-style divisions of labor.

Does Congress Have an Answer?

In February 2025, Senator Dick Durbin introduced the Crypto ATM Fraud Prevention Act, which targets safeguards for kiosk users — warning labels, transaction limits, and operator liability provisions. Wyoming Senator Cynthia Lummis said in September she hopes broader crypto market structure legislation will punish bad actors without stifling innovation.

Both responses assume the problem is solvable with the right guardrails. Maybe. But the machines were designed to be fast and anonymous, and the fraud numbers have climbed every year those design choices remained intact. Call it regulatory optimism — or call it the industry betting it can legislate its way around an infrastructure problem it built itself.