In April 2026, significant DeFi attacks on Drift Protocol and Kelp DAO resulted in nearly $600 million in losses and triggered approximately $9 billion in fund outflows from protocols like Aave. According to Odaily, TRM Labs investigator Nick Carlsen indicated that hackers linked to North Korea may have used AI to assist in target selection and attack path design. Failsafe CEO Aneirin Flynn noted that AI has reduced the time needed to identify blockchain vulnerabilities from months to days or even hours. Reports suggest that Anthropic has not fully released its AI model Mythos due to cybersecurity risks, although the model is said to have the capability to identify large-scale zero-day vulnerabilities. Research indicates that by 2025, over half of blockchain attacks could theoretically be autonomously executed by AI.