AI AGENTS FIND $4.6M IN BLOCKCHAIN SMART CONTRACT EXPLOITS

Summary from the TLDR InfoSec newsletter

SCONE-bench, created by Anthropic researchers, contains 405 exploit-prone smart contracts. Claude Opus 4.5, Sonnet 4.5, and GPT-5 exploited 34 contracts worth $4.6 million. Sonnet 4.5 and GPT-5 found two new zero-day vulnerabilities in 2,849 new contracts, producing exploits worth $3,694 at an API cost of $3,476 for GPT-5. This shows that profitable autonomous exploitation is feasible with current models. Security teams should adopt AI agents for smart contract auditing before deployment, as the decreasing window for exploitation and dropping costs ($1.22 per run) give attackers advantages, while enabling proactive vulnerability detection and fixing.