Anthropic has announced the integration of its specialized cybersecurity model, Mythos, into enterprise security products from Cisco and IBM. The partnership aims to leverage the AI's ability to identify software vulnerabilities and simulate multi-step attacks to help organizations secure their systems before threats are exploited.
Cisco is incorporating Mythos into its AI Defense product to support adaptive red teaming for AI applications and agents. According to Jeff Schultz, senior vice president at Cisco, the tool is designed to probe models for weaknesses, allowing the system to identify and address security threats more effectively.
IBM plans to integrate Mythos into its Concert Protect platform, which focuses on discovering unknown vulnerabilities in customer code and verifying exploitability. While Cisco customers can access the upgrade now, IBM expects to roll out its version in early 2027.
The Mythos model, specifically versions 5.1 and 5.2, is part of Anthropic's Project Glasswing initiative launched in April 2026. To mitigate risks, access is restricted to vetted partners under strict security controls, including mandatory named-operator authentication and comprehensive audit logging.
Anthropic has established a pricing structure for Mythos, charging $10 per million input tokens and $50 per million output tokens. By providing access through established vendors like Cisco and IBM rather than a public API, Anthropic aims to scale its security technology while maintaining oversight.
