Anthropic releases Claude Fable 5 as DeFi hacks top $840M in first five months

Anthropic has released Claude Fable 5, a new AI model it says offers stronger reasoning and coding while blocking the most dangerous uses, as DeFi protocols have lost more than $840 million to hacks in the first five months of the year. According to CoinDesk, Anthropic also made a more capable version, Claude Mythos 5, available only to vetted users in cybersecurity and critical infrastructure, with high-risk requests routed to a weaker Claude Opus 4.8 model in fewer than 5% of sessions.