Cisco page points to six months of change from frontier AI

Cisco is examining how six months of frontier AI have changed the company in a newsroom feature titled “From Flashlight to Floodlight.” The page is dated November 18, 2026, and appears alongside several recent Cisco articles on AI security and agentic AI.
- Cisco’s feature is dated November 18, 2026.
- Its title examines six months of frontier AI at Cisco.
- A listed test reported insecure trust checks 8 of 8 times without learned security rules.
- Cisco’s related articles cover Claude Mythos, AI red teaming and control over AI agents.
A look back at frontier AI
The feature’s title frames Cisco’s experience as a shift from a limited view of frontier AI to a broader one. The source does not provide further details about the changes inside Cisco, specific products, deployment figures or performance results.
Recent focus on AI security
The page lists several related Cisco publications. One article, published November 16, 2026, examines whether a learned security rule changes what an AI assistant writes. Cisco describes a test using the same requests and identical workspaces, assessed by a blind judge. The listing says the agent wrote an insecure trust check “8 of 8 times” without learned security rules, compared with “0 or 1 of 4” times with them.
Another listed article covers Cisco AI Defense adding Anthropic’s Claude Mythos for AI red teaming. The source says Cisco has worked in AI security for several years and describes AI Defense as an enterprise AI security offering that combines asset visibility, supply chain scanning and algorithmic red... The available text ends mid-sentence.
Keeping control of AI agents
The page also links to a November 9, 2026 article titled “Moving Fast and Staying in Control in the Agentic AI Era.” It says the article discusses how handing work to AI agents does not have to mean handing over control. The piece is described as a Cisco Live announcement, with Cisco’s Chief Communications Officer explaining the topic.
What the source supports
Taken together, the listed material shows Cisco focusing its frontier AI discussion on security rules, red teaming and control over AI agents. The supplied source does not establish broader business effects or tell users when any mentioned capabilities will be available.

