Anthropic's bug-finding Mythos model is coming to Cisco and IBM security products

Cisco said at Cisco Live in Melbourne that its AI Defense suite will add Anthropic's Claude Mythos to its adaptive red-teaming tools, letting customers stress-test their own AI apps and agents against registered, authorized targets, at no extra cost under existing licenses. IBM separately said it plans to build Mythos code scanning into Concert Protect to find vulnerabilities traditional scanners miss, with general availability planned for early 2027. Customers get Mythos-powered features through each product's structured interface but no open access to the model, which Anthropic still restricts to vetted organizations.