Anthropic has expanded access to its AI model Claude Mythos, inviting approximately 150 more organizations to join Project Glasswing. The initiative aims to leverage the capabilities of Anthropic’s new frontier model, which has been noted for its potential to reshape cybersecurity.
Originally announced in April, Claude Mythos was released as a preview to select organizations. Anthropic stated that “AI models have reached a level of coding capability where they can surpass all but the most skilled humans at finding and exploiting software vulnerabilities.”
The initial cohort included nearly 50 partners, such as Apple, Amazon, Broadcom, CrowdStrike, Microsoft, and NVIDIA. The new invitees represent more than 15 countries and include industries such as public utilities—specifically power, water, and telecom—along with healthcare providers. Each organization must meet Anthropic’s security requirements to gain access.
Anthropic emphasized the significance of the partnership efforts, noting, “What each partner has in common is that a successful attack on their codebase could be catastrophic.” The company estimates that a major attack could potentially affect over 100 million people, posing serious risks to both global and national security.
Currently, Anthropic is working to safely release Mythos-level capabilities to the public. However, the company has stated that it will not do so until it has implemented “highly robust safeguards” to prevent misuse of its technology.
In May, the unveiling of Claude Mythos influenced the Trump administration’s approach to AI regulation, leading to plans for an executive order aimed at sharing software vulnerability information with critical institutions. That signing was reportedly postponed due to objections from Trump and the tech industry.
