Blockchain intelligence firm TRM Labs measured AI adoption in crypto crime at 54 out of 100 in its 2026 AI-in-Crime Adoption Index—a 40 percent year-over-year increase from approximately 28 in 2024. The index classifies overall AI integration as "emerging," but the underlying breakdown reveals structural risk concentration in fraud.
Scammers show "mature" AI integration. The share of crypto scam reports involving AI—deepfakes, voice synthesis, chatbots—has surged 13 times since 2022. Deepfake scam losses in 2026 have already exceeded the entire 2025 total by 263 percent, according to TRM data.
"AI has not invented new crimes," said Ari Redbord, TRM's global head of policy. "It has removed the constraints on existing ones. The skill floor for cybercriminals has collapsed. The scale ceiling has lifted. Fake identity creation has been industrialized."
The mechanics are brutal in their efficiency. What once required a team of operators now requires one person with a subscription. The barrier to entry for complex attacks has dropped from months of training to hours of setup.
Digital-asset hacks reached 201 in the first half of 2026—more than double the prior-year count. North Korea-linked cyber actors accounted for approximately $600 million in losses, or 61 percent of all reported digital-asset losses in H1 2026. These state-sponsored groups employ deepfake IT-worker infiltration, AI-run social engineering, and AI-assisted vulnerability discovery.
In June, security engineer Taylor Hornby used AI to identify a critical flaw in Zcash's Orchard transaction pool that could have allowed unlimited counterfeit token creation. The vulnerability exemplifies AI's capacity to uncover complex system weaknesses at scale.
TRM also flagged JadePuffer, disclosed as the first fully agentic ransomware attack. An AI agent conducted reconnaissance, stole credentials, executed lateral movement, and performed encryption without human intervention. "Agentic attacks represent the future model for large-scale assaults against critical infrastructure, including hospital systems," Redbord said. "They require no human in the loop. The operational efficiency is rare. The potential scale is civilization-level."
No-code ransomware kits now cost $400 to $1,200, drastically lowering entry barriers. Combined with illicit value flowing openly on public blockchains, this elevates risk across cryptocurrency exchanges, DeFi protocols, DEXs, and token launches.
TRM's index shows uneven penetration: scams at "mature" integration, hacking and ransomware at "emerging," narcotics and darknet markets at "horizon." The 40 percent increase in overall AI adoption correlates directly with the 263 percent spike in deepfake losses and the doubling of digital-asset hacks. The data suggests this structural shift—lower skill floors, amplified scale, automation of reconnaissance-to-execution—is durable.
