Anthropic confirms Claude Mythos-class models will roll out to the public

Anthropic plans to roll out Mythos-class AI models to the public after delaying the launch to build stronger safeguards against misuse; the company says Mythos offers major improvements in code reasoning and autonomy and has been previewed to a small set of organizations (including for cybersecurity work). The report emphasizes security concerns about powerful models and Anthropic’s assertion that robust guardrails are needed to prevent attackers from gaining an advantage.