AI-Driven Attacks Escalate DeFi Losses to Record Levels

The rise of artificial intelligence (AI) in cyber attacks has been a growing concern in the decentralized finance (DeFi) sector. Recent research has revealed that AI agents can autonomously discover and exploit vulnerabilities in DeFi protocols, with a success rate of up to 70% when given access to structured knowledge about common attack patterns.

GoPlus Security reported four separate smart contract exploits on Ethereum mainnet within a 48-hour window ending April 29, resulting in combined losses exceeding $1.5 million. The firm warned that hackers armed with AI are becoming more precise and faster than ever.

a16z crypto tested an off-the-shelf AI coding agent against 20 past price manipulation incidents on Ethereum and found a success rate of 10% without access to structured knowledge. However, when given this information, the success rate jumped to 70%. The researchers noted that while the AI is effective at finding bugs, it sometimes struggles with complex, multi-step attacks.

Anthropic recently announced a new AI model called Claude Mythos Preview, which can autonomously find and write working exploits for zero-day vulnerabilities across major operating systems and web browsers. The company confirmed that the same improvements that make the model good at patching vulnerabilities also make it good at exploiting them.