Anthropic: 'Autonomous Exploitation' of Blockchain Contracts for Profit is Technically Feasible

Researchers at Anthropic demonstrated that frontier models like Claude Opus 4.5 and GPT-5 can autonomously identify and exploit zero-day vulnerabilities in smart contracts. In simulations, these agents generated over $4.6 million in "profit," proving that profitable, real-world autonomous exploitation is now a technical reality.