Expanding Cybersecurity Capabilities of Claude Mythos 5 to More Defenders

Anthropic announced updates to broaden the use of its top-tier frontier model, Claude Mythos 5, for cyber defense. The core of the update is expanding the scope of 'Project Glasswing', which was previously limited to select organizations, to increase defensive access through safety classifiers and safeguards.

Key highlights include:

• Partner Tool Integration: Integrating Claude Mythos 5 into the products and services of cybersecurity partners, minimizing risk by providing users with specific outputs such as patch suggestions or security alerts without direct access to the model.
• Claude Security Update: Claude Enterprise customers can now use Claude Mythos 5 for Claude Security scans, billed at standard token usage rates within existing plans.
• Defender Advantage Fund (0xDAF): Establishing a new fund providing $35 million in credits to organizations supporting vulnerability patching and security automation processes for open-source projects.
• Cyber Verification Program Expansion: Allowing verified defenders access to dual-use capabilities of the Opus and Sonnet models, with plans to include access to Mythos-class models in the future.