Anthropic opens the door on Mythos cybersecurity findings

Anthropic is letting users of its Mythos artificial-intelligence model share cybersecurity threats with other organisations that may face similar vulnerabilities.

Anthropic, the AI company, modified its previous stance after concerns that limiting access to that information could hurt smaller companies, the report said.

The change follows debates inside the industry about how tightly firms should control access to their most capable models, and what those controls mean for defenders who rely on timely threat information.

The move highlights the practical trade-offs AI companies confront when balancing model safety, intellectual-property protection and the operational needs of customers defending against fast-moving cyberattacks.