Anthropic announced Tuesday that it will expand access to its Claude Mythos cybersecurity model to about 200 Project Glasswing partners [1].
This rollout marks a significant scaling of the company's specialized security tools. By broadening the user base, Anthropic aims to meet rising demand for advanced AI security and strengthen its overall cybersecurity ecosystem [1], [2].
The expansion represents a four-fold increase from the previous group of more than 40 partners [2], [3]. As part of this latest phase, the company is adding 150 new partners [4]. These new collaborators span more than 15 countries worldwide [4].
Based in San Francisco, Anthropic has previously limited the scope of the Mythos model. Prior to this announcement, access was restricted to the U.S. government and a small circle of Project Glasswing partners [2].
The company said the move is intended to broaden the deployment of its security capabilities. The initiative focuses on deploying AI tools that can better defend against complex digital threats across various international jurisdictions [1], [2].
“Anthropic is quadrupling the number of Project Glasswing partners.”
The transition of Claude Mythos from a closed government-centric tool to a broader partner network suggests Anthropic is moving toward a more commercialized or collaborative security model. By diversifying its user base across 15 countries, the company can gather more diverse telemetry on global cyber threats, potentially accelerating the model's ability to detect and mitigate emerging vulnerabilities in real-time.
