AI models led to a ‘vulnerability apocalypse’ in crypto security: Immunefi CEO

Immunefi CEO Mitchell Amador has linked the recent surge in decentralized finance (DeFi) hacks to the rapid advancement of AI models, including Claude Opus 4.8 and ChatGPT 5.5. Speaking at the WAIB Summit in Monaco, Amador described the situation as a 'vulnerability apocalypse,' where attackers now leverage AI to exploit weaknesses in crypto protocols. April 2026 saw illicit actors steal over $634 million, the highest monthly total since the Bybit hack contributed to $1.4 billion in losses in February 2025, according to DefiLlama. Amador emphasized that the next three to four years will be pivotal for the crypto industry’s survival, as defenders work to harness AI for creating 'impregnable' codebases. He suggested this timeline could shorten to under two years if crowdsourced security solutions are adopted more widely. His remarks followed Anthropic’s release of its Claude Mythos model, Fable 5, which raised concerns despite safeguards rerouting cybersecurity-related queries to Claude Opus 4.8. The industry’s security concerns intensified after a series of major exploits, including the Kelp DAO hack on April 19. Attackers exploited a single point of failure in LayerZero’s 1/1 decentralized verifier network (DVN) setup, draining approximately 116,500 restaked Ether (rsETH), valued at $290 million to $293 million. LayerZero had previously advised against this configuration, acknowledging it created vulnerabilities in cross-chain messaging. Amador’s warnings highlight the urgent need for proactive measures to counter AI-driven threats. While attackers currently exploit these tools, the crypto sector must adapt by integrating AI into defensive strategies. The Kelp DAO incident underscores the risks of protocol misconfigurations, reinforcing the call for stronger security frameworks in DeFi.