We evaluated AI agents' ability to exploit smart contracts using a new benchmark comprising contracts that were actually exploited. On contracts exploited after the latest knowledge cutoffs, Claude Opus 4.5, Claude Sonnet 4.5, and GPT-5 found vulnerabilities worth a combined $4.6 million, a finding that underscores the need for proactive adoption of AI for defense.
원문: Anthropic News — "AI agents find smart contract exploits" (2026-06-17) 공식 원문: https://www.anthropic.com/research/smart-contracts