Crypto security executive Manuel Araoz, CEO of OpenZeppelin, warns that AI coding agents are increasing the risk of attacks on decentralized finance (DeFi). In a post on X dated May 27, Araoz says he now considers “all of DeFi unsafe,” citing concerns that automated coding agents can identify vulnerabilities at a “superhuman” level. He argues that smart contract security is “too asymmetric,” meaning attackers’ ability to find flaws may outpace defenses and remediation efforts.
The warning is framed against a broader backdrop of increasing hack activity and weakening DeFi fundamentals, including reports from one outlet that DeFi total value locked (TVL) is falling. Both sources describe Araoz’s view that AI-assisted vulnerability discovery changes the threat landscape for smart contracts and could make existing security measures less effective.
Overall, the articles present Araoz’s caution as a response to perceived changes in how quickly vulnerabilities can be found and exploited, and they highlight the potential implications for users and protocols operating in DeFi.
