Cisco says it will add Anthropic’s Claude Mythos model to its AI Defense product to support adaptive red teaming of AI applications and agents, as the company positions for wider enterprise adoption of AI agents that can access tools and sensitive data.
Announced at Cisco Live APJC in Melbourne, the update is part of Cisco’s partnership with Anthropic and is intended to help security teams simulate multi-step attacks against AI agents and identify weaknesses before systems are deployed to production.
“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 said Claude Mythos will augment its AI Defense red teaming framework by generating contextual attack paths, translating test results into follow-up attack ideas for each target system, and recording actions for audit. Cisco also said customers will interact with the model through AI Defense’s interface rather than having open-ended prompt access.
The company said it has built safeguards into the integration, including limiting red teaming to explicitly registered and authorised targets owned by the enterprise customer, requiring authentication tied to a named operator, and supporting time-bounded engagements that can be halted on demand.
The announcement also includes updates to Splunk Agent Observability aimed at linking AI agent cost with performance and business impact. Cisco said its “Tokenomics” capability will use usage data from Amazon Bedrock and Amazon CloudWatch to attribute costs to specific agents and pair that with evaluations that grade the quality and impact of the agents’ work. Cisco said Tokenomics will be available in private preview in December, with general availability in early 2027 via AWS Marketplace.
Cisco also announced generally available updates to Splunk Agent Observability evaluation capabilities, including Eval-G1 evaluation models and an “evaluation tuner” designed to help teams create fine-tuned evaluator models from traces and subject matter expert feedback, with the aim of enforcing quality standards and controlling agents in near real time.
Separately, Cisco said it is enhancing Cisco IQ to correlate live threat detection data with vulnerability and lifecycle information about network infrastructure, using integrations intended to bring that context into security workflows. Cisco said its Cisco Data Fabric will ingest Cisco IQ vulnerability, lifecycle and configuration data into the Splunk platform via open APIs so security operations teams can correlate threat detections with exposure information in Splunk Enterprise Security.
Other Cisco IQ updates announced include Model Context Protocol and ServiceNow connectors to translate exposure findings into asset-linked ServiceNow problem records, and an AI-powered resilient infrastructure playbook intended to surface lifecycle vulnerabilities, defect alerts and “quantum-safe cryptographic readiness” information for planning modernisation roadmaps.
“AI is shrinking the time organizations have to respond to threats, and knowing where you’re exposed only matters if you can act on it,” said Liz Centoni, Cisco EVP and Chief Customer Experience Officer. “Bringing Cisco IQ vulnerability insights into Cisco Data Fabric powered by the Splunk Platform gives analysts that context in the middle of an investigation, so they can decide what to do next. With our partners and Professional Services teams, we’re helping customers move from identifying risk to actually reducing it.”
Cisco cited a Splunk research report stating that 81% of ITOps and engineering professionals reported increased cooperation between security and observability teams as AI agents proliferate. Cisco said it has supported hundreds of customers with AI Defense since the product’s launch in January 2025.
Cisco said adaptive red teaming with Claude Mythos is “coming soon” for select enterprises.
