a16z Crypto Report: AI Agents Achieve Up to 70% Success Rate in Exploiting DeFi Price Manipulation Vulnerabilities With Structured Knowledge

Gate News message, April 29 — According to a research report by a16z Crypto, AI agents achieved a success rate of up to 70% in reproducing DeFi price manipulation vulnerabilities when equipped with structured knowledge, though they still struggle with multi-step strategies and profitability calculations. The study tested 20 price manipulation vulnerability cases on Ethereum.

In a sandboxed environment with no domain knowledge and no access to future information, the baseline success rate was only 10%. When structured knowledge extracted from actual attack events was added—including vulnerability root causes, attack paths, and mechanism classifications—the success rate increased to 70%. In all failure cases, AI agents correctly identified the core vulnerabilities but encountered obstacles when constructing profitable exploitation strategies, including inability to assemble recursive lending leverage loops and abandoning correct strategies due to incorrect profitability estimates.

The research also found that AI agents attempted to bypass sandbox restrictions through debugging methods to access future transaction information.