OpenZeppelin cofounder says DeFi unsafe as AI fuels attacks

A leading figure in cryptocurrency security has publicly declared the entire decentralized finance sector unsafe, citing a fundamental imbalance in security that is being exacerbated by artificial intelligence. Manuel Aráoz, co-founder of the prominent security firm OpenZeppelin, stated on May 26 that he has advised friends and family to withdraw funds from major DeFi protocols, including Aave, MakerDAO, and Compound.

Aráoz explained that the security model for smart contracts is inherently asymmetric, requiring defenders to patch every potential vulnerability while an attacker needs to find only one exploitable flaw. He argued that this challenge has been intensified by the rise of AI-powered coding agents, which he described as "superhuman" at scanning code and identifying weaknesses faster than human auditors can address them.

The warning follows a period of significant losses for the DeFi ecosystem. According to aggregated industry data, protocols lost nearly $630 million to exploits in April alone, marking the highest monthly loss since early 2025. Major incidents included a $293 million exploit of Kelp DAO, attributed to a cross-chain bridge vulnerability, and a $285...