Palo Alto Networks launches continuous AI penetration-testing service

Palo Alto Networks launched Continuous Frontier AI Defense, a worldwide subscription service that repeatedly tests customers' web applications, APIs, cloud infrastructure, code repositories and networks. Unit 42 software routes tasks among Anthropic's Claude Mythos 5, OpenAI's GPT-5.6-Cyber and open-weight models, then attempts to verify vulnerabilities, map possible attack paths and propose fixes or temporary virtual patches. The service extends an earlier one-time assessment. Palo Alto Networks says six months of development across more than 100 engagements found exposures at every tested customer, with 37% rated high or critical; those results are company-reported.