{"id":307,"date":"2026-11-11T11:06:19","date_gmt":"2026-11-11T10:06:19","guid":{"rendered":"https:\/\/mspbusiness.com\/en\/it-news\/cisco-integrates-anthropics-claude-mythos-into-ai-defense\/"},"modified":"2026-11-11T17:09:16","modified_gmt":"2026-11-11T16:09:16","slug":"cisco-integrates-anthropics-claude-mythos-into-ai-defense","status":"publish","type":"post","link":"https:\/\/mspbusiness.com\/en\/technology-and-architecture\/cisco-integrates-anthropics-claude-mythos-into-ai-defense\/","title":{"rendered":"Cisco integrates Anthropic’s Claude Mythos into AI Defense"},"content":{"rendered":"

Cisco is adding Anthropic’s Claude Mythos to AI Defense, enabling organisations to test their AI applications and agents for vulnerabilities. The company announced this at Cisco Live APJC in Melbourne. Cisco is also expanding Splunk Agent Observability with Tokenomics and evaluation models, and linking Cisco IQ to security workflows.<\/strong><\/p>\n

The addition allows organisations to carry out red teaming on their AI systems using a frontier AI model, Cisco says. Jeetu Patel, President and Chief Product Officer at Cisco, says organisations can only scale AI if they can secure, manage and observe it. By combining runtime protection with quality controls for agents, Cisco aims to prevent agents from being compromised. The company also wants to show whether every token used delivers measurable business value.<\/p>\n

Red teaming with Claude Mythos<\/h3>\n

Cisco says that since launching AI Defense in January 2025 it has helped hundreds of customers detect vulnerabilities in AI models, applications and agents before they go into production. According to the company, Claude Mythos adds reasoning capabilities to this. The model plans tests and draws up multi-step attack paths for each system in order to find subtle vulnerabilities. Every test action is logged for auditing. Customers work exclusively through the structured interface of AI Defense and are not given open prompt access.<\/p>\n

Michael Moore, Cybersecurity Lead at Anthropic, points out that organisations are increasingly deploying AI agents for their core processes. According to Moore, thorough testing before agents go into production is therefore essential.<\/p>\n

Restrictions on the model<\/h3>\n

AI Defense imposes several safeguards on the use of Claude Mythos. The model only operates on systems that the customer has registered, owns and has approved for testing. Every task requires authentication and is cryptographically linked to a named operator within the organisation. A human user determines the task and the objectives. Sessions are time-limited and can be stopped at any moment. In addition, the controls in AI Defense work alongside programme-level safeguards and monitoring.<\/p>\n

Tokenomics and evaluation models in Splunk<\/h3>\n

Splunk Agent Observability is gaining a feature called Tokenomics, which, according to Cisco, links the costs of each agent to evaluations of the quality and impact of its work. To do so, Tokenomics uses usage data from Amazon Bedrock and Amazon CloudWatch and shows the ROI of AI deployment in a single overview.<\/p>\n

The company is also introducing Eval-G1 Evaluation Models and The Evaluation Tuner. These allow teams to turn agentic traces and expert feedback into their own small language models for evaluations and guardrails. Cisco says these models are up to 98 per cent cheaper to run than a commercial LLM and respond 2.5 times faster than traditional LLMs.<\/p>\n

Cisco IQ in security workflows<\/h3>\n

Cisco is bringing real-time intelligence from Cisco IQ directly into security environments and automated workflows in platforms including Splunk and ServiceNow. Cisco IQ consolidates knowledge about vulnerabilities and, according to the company, links it in real time to threats on the network. The platform is used by more than 16,000 customers with nearly 20 million devices between them. Cisco justifies the move by arguing that AI is shortening the time between the disclosure and the exploitation of a vulnerability from weeks to hours.<\/p>\n

Liz Centoni, EVP and Chief Customer Experience Officer at Cisco, says insight into vulnerabilities is only useful if organisations can also act on it. According to Centoni, Cisco supports customers in this together with partners and its Professional Services teams.<\/p>\n

Availability<\/h3>\n

Adaptive red teaming with Claude Mythos will soon be available to selected organisations. Tokenomics is expected to launch in December as a private preview, with general availability in AWS Marketplace in early 2027. The evaluation models and the Evaluation Tuner are generally available from this month.<\/p>\n","protected":false},"excerpt":{"rendered":"

Organisations can test AI applications and agents in Cisco AI Defense using frontier model Claude Mythos. Splunk Agent Observability gains Tokenomics, linking agent costs to quality.<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"mspb_bron_id":1169646,"mspb_url_nl":"https:\/\/mspbusiness.com\/technologie-en-architectuur\/cisco-integreert-claude-mythos-van-anthropic-in-ai-defense\/","mspb_url_de":"https:\/\/mspbusiness.com\/de\/technologie-und-architektur\/cisco-integriert-claude-mythos-von-anthropic-in-ai-defense\/","mspb_url_en":"","mspb_beeld_url":"https:\/\/mspbusiness.com\/wp-content\/uploads\/2026\/11\/hero-1794391565535.png","mspb_beeld_klein":"https:\/\/mspbusiness.com\/wp-content\/uploads\/2026\/11\/hero-1794391565535.png","mspb_auteur":"Redactie","mspb_beeld_alt":"","mspb_video":"","_yoast_wpseo_metadesc":"Cisco adds Anthropic's Claude Mythos to AI Defense for red teaming and expands Splunk Agent Observability with Tokenomics.","footnotes":""},"categories":[5],"tags":[],"layout":[6],"class_list":["post-307","post","type-post","status-publish","format-standard","hentry","category-technology-and-architecture","layout-news"],"_links":{"self":[{"href":"https:\/\/mspbusiness.com\/en\/wp-json\/wp\/v2\/posts\/307","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/mspbusiness.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/mspbusiness.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/mspbusiness.com\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/mspbusiness.com\/en\/wp-json\/wp\/v2\/comments?post=307"}],"version-history":[{"count":1,"href":"https:\/\/mspbusiness.com\/en\/wp-json\/wp\/v2\/posts\/307\/revisions"}],"predecessor-version":[{"id":308,"href":"https:\/\/mspbusiness.com\/en\/wp-json\/wp\/v2\/posts\/307\/revisions\/308"}],"wp:attachment":[{"href":"https:\/\/mspbusiness.com\/en\/wp-json\/wp\/v2\/media?parent=307"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/mspbusiness.com\/en\/wp-json\/wp\/v2\/categories?post=307"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/mspbusiness.com\/en\/wp-json\/wp\/v2\/tags?post=307"},{"taxonomy":"layout","embeddable":true,"href":"https:\/\/mspbusiness.com\/en\/wp-json\/wp\/v2\/layout?post=307"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}