Anthropic Releases AI Tool It Called Too Dangerous

Anthropic has released Claude Fable 5, a version of its powerful Mythos AI model, to the public after months of restricted testing. The company acknowledged risks but said safeguards and user limitations are in place. Mythos previously raised concerns among government and finance leaders over its ability to exploit computer systems. Around 150 organisations that previewed Mythos have reported over 10,000 critical security flaws found using the tool. A less restricted version, Claude Mythos 5, is also being expanded to cybersecurity groups. Anthropic co-founder Jack Clark has warned the AI industry lacks a brake pedal as capabilities rapidly advance.