Anthropic’s new AI model accelerates vulnerability detection in crypto systems, potentially worsening $840 million in DeFi losses this year.
Anthropic released Claude Fable 5, an AI model with enhanced reasoning and coding capabilities, designed to identify security flaws at superhuman speeds. The tool could amplify risks in decentralized finance (DeFi), which has already suffered over $840 million in hacks in 2024.
DeFi breaches this year have largely resulted from human error and operational failures, not smart-contract bugs. Security experts warn that while AI won’t invent new attack vectors, it will expedite exploitation of misconfigurations, exposed keys, and flawed signing processes.
The public version of Claude Fable 5 includes safety filters to block malicious use, while a restricted Mythos 5 variant offers unrestricted access to vetted users. The release arrives as crypto markets face persistent security challenges.