Manuel Aráoz, co-founder of OpenZeppelin, has issued a stark warning about the safety of decentralized finance (DeFi) protocols, citing the rise of AI-powered coding agents. Aráoz advised against trusting any DeFi platforms, including major ones like Aave, MakerDAO, and Compound, due to the enhanced capabilities of AI in identifying vulnerabilities faster than human teams. He highlighted the asymmetry in security, where defenders must address all vulnerabilities while attackers need only exploit one. Despite Aráoz's concerns, some industry figures have pushed back. Mark Zeller of Aave Chan Initiative argued that less than 10% of DeFi issues stem from code vulnerabilities, attributing most problems to poor risk management and operational security. Others, like Phoenix Lab's Sam McPherson, emphasized that smart contracts are generally secure, with recent breaches often involving centralized components. Ethereum co-founder Vitalik Buterin suggested AI could enhance security through formal verification, potentially strengthening crypto systems over time.