Anthropic has built one of its most powerful AI systems yet and chosen not to release it broadly.
Its unreleased model, Claude Mythos Preview, reportedly identified thousands of high-severity software vulnerabilities across major operating systems and browsers. Instead of public release, Anthropic is limiting access to selected organisations focused on defensive cybersecurity.
To support that effort, it has launched Project Glasswing with partners including Amazon Web Services, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorgan Chase, the Linux Foundation, Microsoft, Nvidia and Palo Alto Networks.
This is what makes the story stand out. It is not only about capability. It is also about restraint, accountability and the growing importance of AI governance.
For business leaders, this brings a practical question into focus. As AI becomes more embedded in systems and workflows, what framework do you have in place to define what it should and should not be allowed to do?
📰 Sources:
