Cisco adds Anthropic’s Claude Mythos to AI Defense red teaming

Mythos joins Cisco’s adaptive testing engine

Cisco revealed the integration on Monday at Cisco Live APJC in Melbourne. AI Defense will combine Claude Mythos with safeguards that include testing only authorized targets, linking each engagement to a named operator and providing full auditability.

According to Cisco, the controls are intended to let red team exercises mirror real adversaries without endangering customer environments. DJ Sampath, senior vice president and general manager of Cisco’s AI Software and Platform group, said organizations need to understand their specific use cases before testing them for security gaps.

“You need to really deeply understand each of [the] use cases so that you’re red teaming against those use cases to be able to discover those gaps,” Sampath told CRN. “We’re moving from generic red teaming to adaptive red teaming, but more importantly, we’re using Mythos model to be able to do it.”

From common vulnerabilities to tailored attacks

AI Defense, first introduced in 2025, helps enterprises address misuse of AI tools and data leakage, while protecting AI models, applications and agents. It already offers standard and adaptive red teaming.

Standard tests cover thousands of common vulnerabilities, including prompt injection, jailbreaks and sensitive data exposure. The tool can generate reports mapped to Cisco’s AI security framework and other security standards.

Adaptive red teaming allows users to define custom criteria in plain language. The framework then plans and runs multi-step attacks tailored to the applications being tested. Cisco says Claude Mythos adds attacker-level reasoning to this process, with safeguards intended for authorized testing.

Cisco is running Mythos in a sandbox environment with control over what it can access, how it executes and how its sub-agents are created. Sampath said this visibility also supports “tokenomics”, allowing organizations to examine token use from both security and cost perspectives.

A partner opportunity around agent security

Cisco partners can use Mythos through AI Defense to red team client environments where agents are being deployed, Sampath said. He described the capability as “a game changer for [partners]” and said he knew of no other red teaming service with a model as powerful.

AI Defense is part of Cisco Cloud Control, the company’s unified operations platform for AgenticOps, where human operators and AI agents manage IT infrastructure together. Cisco also said AI Defense now supports Amazon Bedrock and Amazon CloudWatch, allowing partners and users to view token spend and return on investment in one dashboard.

Cisco said AI Defense has approximately 400 to 500 customers since its launch last year. Adaptive red teaming with Claude Mythos is coming soon for select enterprises. Broader availability is planned for qualified customers, although dates, scope and regional availability remain subject to change.