The crypto community is concerned that the security mechanisms of Anthropic's Claude Fable 5 could be exploited by hackers.

According to a report by Cointelegraph, Anthropic has released the first public version of its Claude Mythos model, Fable 5. Despite built-in security guards, crypto users are concerned that it could be used for malicious purposes. The model has previously been found to have over 10,000 high-risk vulnerabilities in "systemically important software." Anthropic stated that Fable 5 is security-ready for general use, and the guards will redirect discussions on cybersecurity to other models. Simon Dedic, founder of Moonrock Capital, warned that the cost and skill requirements for finding smart contract vulnerabilities using Fable 5 will be "virtually zero," making unaudited protocols easy targets and allowing known vulnerabilities to be repeatedly exploited. He advised users to revoke wallet authorization, remove assets from the protocol, and transfer them to a new hardware wallet. Michael Egorov, co-founder of Curve, believes the threat is exaggerated, noting that smart contract code is only a few thousand lines long, which humans and regular AI can already reason about effectively; the real risks lie in operational security and supply chain attacks.