Anthropic released two new AI models Monday: Fable 5.2 and Mythos 5.2. CNET's early read is that the pair is overkill for average users but meaningfully smarter for professional workloads — a positioning consistent with the industry's drift toward segmenting frontier models by use case rather than chasing a single flagship.

The enterprise angle arrived in parallel. Cisco will integrate Claude Mythos into its AI Defense product for red teaming — using the model to plan attacks against customers' own AI applications and agents to surface vulnerabilities, The Next Web reports.

The guardrails are notable: customers get no open prompt access to the model, and every test is logged. That design acknowledges the dual-use problem at the heart of offensive security tooling — the same capabilities that probe weaknesses could be misused without tight controls.

Taken together, the day's news shows Anthropic pushing on two fronts at once: raw capability for professionals, and embedded, constrained deployments inside enterprise security products where the model operates as infrastructure rather than chatbot.