TRM Labs: AI-Assisted Crypto Crime Surges 40% as Scams Hit 54/100

AI Adoption in Crypto Crime Climbs 40%

Blockchain intelligence firm TRM Labs reports that AI adoption across cryptocurrency-related crime has climbed 40% over the past year, with scam operations leading the charge. The latest edition of TRM Labs' AI Adoption Index for the crypto crime landscape assigns an overall score of 54 out of 100 for AI usage in the sector, up from 28 in 2024 and now classified as “emerging.” Fraud schemes have already reached the “mature” stage, while hacks and ransomware remain in the emerging category. According to TRM Labs, AI has not created entirely new crime types, but has removed long-standing technical barriers and allowed attackers to scale operations.

In a related trend, data from venture capital firm Andreessen Horowitz (a16z) shows that AI agents are transforming legitimate crypto-adjacent industries as well. The firm's latest Charts of the Week, citing OpenRouter data, indicates that agents now consume nearly five times more tokens than human users, with usage climbing roughly 14-fold since February. a16z links the shift to rising demand for high-bandwidth memory, because agents rely heavily on cached tokens rather than simple prompt-response interactions. The report also notes that the top 10% of enterprise users by token output have extended their lead over typical firms to about eight times.

Alongside the productivity narrative, capital flows show a structural rotation that could shape crypto and tech markets. a16z's analysis describes a broader shift from “bits” to “atoms,” with ETF themes moving from clean energy and healthcare toward AI, nuclear energy and space. Data centers are paying a wage premium of roughly 50% for skilled blue-collar workers, according to Indeed and ADP figures cited in the report. That shift is already visible in hiring patterns, with construction and skilled trades seeing stronger job-switching wage growth, while traditional automation tools such as Zapier have experienced double-digit traffic declines.

The danger posed by AI in crypto has also produced concrete losses. TRM Labs notes that reports of deepfake-related or AI chatbot-driven crypto fraud have increased 13-fold since 2022, and financial damage from deepfake scams so far in 2026 already exceeds the full-year 2025 total by 263%. North Korean hacker groups have been using deepfakes to pose as IT workers, while a security researcher discovered a critical vulnerability in Zcash's Orchard transaction pool using AI earlier this year. Hacking incidents reached a record 201 in the first half of 2026, more than double the year-earlier total, with roughly 61% of stolen funds linked to North Korean actors.

Regulatory and Market Implications

Together, these separate data points from TRM Labs and a16z point to a single theme: AI is becoming a decisive variable in both crypto crime and crypto adoption. The TRM Labs index, which is based on criminal cases and fraud reports, demonstrates that the barrier to launching sophisticated attacks has fallen dramatically. As the legal sector's Codex adoption surged 108-fold since February and enterprise AI usage diverges sharply between leaders and laggards, the same tools are being weaponized by bad actors. The data suggests that security teams and regulators must treat AI-augmented fraud not as a future risk but as the current baseline.