Anthropic halts public rollout of AI tool over cybersecurity fears

Anthropic has confirmed it will not release its advanced AI model, Claude Mythos, to the public due to its powerful ability to uncover software vulnerabilities. The model has exposed thousands of flaws in widely-used applications, including some dating back nearly three decades, none of which had been detected previously. Fearing misuse by hackers, the San Francisco-based startup has partnered with cybersecurity firms and tech giants like Amazon, Apple, Microsoft, and Cisco as part of a collabor