OpenZeppelin Ex-CTO Warns AI Makes DeFi Unsafe Amid $1.1B Exploit Losses

Manuel Aráoz, co-founder and former chief technology officer of OpenZeppelin, warned that artificial intelligence (AI) developments are undermining decentralised finance (DeFi) security, claiming "all" of DeFi is now unsafe. Aráoz stated AI coding agents have become exceptionally effective at identifying weaknesses in publicly available smart contract code, creating a growing imbalance where developers must patch every vulnerability while AI-assisted attackers only need to locate a single flaw. The warning comes as more than US$1.1 billion (AU$1.54 billion) has been lost to DeFi exploits over the past 365 days, according to DefiLlama figures, raising concerns about whether DeFi's security framework remains viable against AI-driven threats.

DeFi Exploit Losses in April 2026

April 2026 recorded more than US$600 million (AU$840 million) in protocol losses, according to DefiLlama. The month's largest exploits included US$292 million (AU$408.8 million) from KelpDAO, US$285 million (AU$399 million) from Drift, and US$197 million (AU$275.8 million) from Euler. Anthropic warned that its restricted Claude Mythos AI model can independently discover software vulnerabilities and produce working exploits at a level surpassing existing automated systems.

OpenZeppelin Response

OpenZeppelin stated that Aráoz's views do not reflect the company's current position. On May 27, 2026, the company posted on Twitter: "We have secured DeFi for a decade, and that work now matters more than ever. We are in it alongside the protocols, institutions, and developers building the next era of finance."