Cisco announced it is adding Claude Mythos to AI Defense for red teaming AI applications and agents, a move the company says will let defenders stress-test AI systems with frontier adversary reasoning. The announcement was made at Cisco Live APJC in Melbourne, Australia, and pairs the red teaming addition with new Splunk Agent Observability capabilities and Cisco IQ advancements.

Cisco said the combined innovations are intended to prepare enterprises to secure, observe, and govern agents at machine speed.

A Continuous Trust Loop

“You cannot scale what you cannot secure, govern, or observe,” said Cisco President and Chief Product Officer Jeetu Patel. "By uniting frontier-level AI runtime protections with machine-speed agent quality controls, we are building a continuous trust loop: protecting agents from compromise while ensuring every token spent delivers measurable business value."

Cisco cited a global Splunk research report from last month finding that 81% of ITOps and engineering professionals cite increased cooperation between security and observability teams, as AI agents proliferate and connect to complex tools and sensitive data.

Cisco said its Splunk and AI Defense innovations equip security leaders to account for token spend and effective tool use while conducting tailored security testing against highly capable AI agents.

AI Defense and Claude Mythos

Since Cisco launched AI Defense in January 2025, the company said it has helped hundreds of customers identify critical flaws in AI models, applications, and agents before they reach production. Its adaptive red teaming allows security teams to build customized, multi-step attacks on target systems to uncover specific risks.

Cisco said Claude Mythos will augment this framework with advanced reasoning that plans tests and translates results into attack ideas for each system.

Features Cisco listed include:

  • Realistic Adversarial Ideation: generates multi-step, contextual attack paths designed to help identify subtle vulnerabilities across complex architectures.
  • Secure, Contained Access: customers interact only through AI Defense's structured interface, with no open-ended prompt access.
  • Audit Trails and Repeatable Results: AI Defense records each testing action for audit, and results can be carried over to inform follow-up attack loops.
"Enterprises are increasingly building AI agents into their core operations and want to test those agents thoroughly before putting them into production,“ said Michael Moore, Cybersecurity Lead at Anthropic. ”By bringing Claude Mythos to Cisco AI Defense, Cisco can help security teams red team their agents with advanced reasoning that plans tests and generates attack ideas for each system."

Safeguards Around Red Teaming

Cisco said it engineered AI Defense to use Claude Mythos with multi-layered safeguards:

  • Registered, Authorized Targets: operates exclusively against targets explicitly registered, owned, and authorized for testing by the enterprise customer, and cannot scan arbitrary internet infrastructure.
  • Named Operator Attribution: every red teaming engagement requires authentication and is cryptographically tied to a named enterprise operator.
  • Revocable Execution: engagements run with a fixed job and objectives defined by the human user, with time-bound sessions that can be halted on demand.
  • Dual-Layer Oversight: product-level controls in AI Defense operate alongside program-level safeguards and monitoring.

From Test Findings to Measurable Agent ROI

Cisco said frontier AI models can excel at complex security tasks such as crafting AI red teaming strategies, but that the capability cuts both ways, as AI lowers the barrier to entry for attackers while enterprises work to balance security, governance, and ROI across growing AI portfolios.

Cisco said organizations need unified, real-time visibility into AI agent cost, business performance, and quality to deploy agents at scale. It pointed to new Splunk Agent Observability enhancements:

  • Tokenomics: cost governance, soon to be available in AWS Marketplace, that ties AI spend to output quality. Using usage data from Amazon Bedrock and Amazon CloudWatch, it pairs each agent's cost with evaluations that grade the quality and impact of its work, giving organizations a single view of AI ROI. Cisco said this extends the token-spend attribution launched at .conf26, which tracks costs across AI agents and coding tools such as Claude Code, Codex, and Cursor.
  • Eval-G1 Evaluation Models: Splunk Agent Observability extends support for turning agentic traces and expert feedback into custom small language models that execute evals and guardrails at up to 98% lower cost than a commercial LLM. Cisco said these next-generation Eval-G1 models deliver evaluation response times 2.5x faster than traditional LLMs and enable real-time agent control with high accuracy.
  • The Evaluation Tuner: teams can turn their agentic traces and subject matter expert feedback into fine-tuned evaluator models to enforce quality standards and steer agents at machine speed.

Availability

Cisco said adaptive red teaming with Claude Mythos is coming soon for select enterprises. Tokenomics will be available in private preview in December, with general availability in AWS Marketplace early 2027. The evaluation models and the evaluation tuner are generally available starting this month.

Cisco IQ and Data Fabric

Cisco said Cisco IQ, which leverages decades of vulnerability and troubleshooting data, now correlates live threat detection data with potentially exposed network infrastructure at machine speed to accelerate remediation. The company framed the Cisco IQ and Cisco Data Fabric work as connecting infrastructure exposure to security action.