The cybersecurity company's newest product runs Anthropic's and OpenAI's models side by side against the same target: the corporate networks hackers are using AI of their own to break into.
Palo Alto Networks this week said it would launch a new cybersecurity service, Unit 42 Continuous Frontier AI Defense, that uses AI models from both Anthropic and OpenAI, alongside open-weight models, to continuously test corporate networks for vulnerabilities.
The service is designed to probe web applications, application programming interfaces and cloud infrastructure on an ongoing basis, identifying vulnerabilities and possible attack paths as a company's digital environment changes, rather than relying solely on periodic manual security audits. Beyond detection, the offering is meant to provide guidance on fixing identified gaps, including code-level fixes and virtual patching options that can shore up a vulnerability before a permanent fix is deployed.
The product will be available globally through annual subscriptions, with pricing based on which mix of Anthropic, OpenAI and open-source models a customer selects, giving buyers some flexibility over cost and model choice within a single security offering.
The launch reflects a broader shift in enterprise cybersecurity: attackers are increasingly using AI to find and exploit weaknesses faster than manual security teams can respond, pushing defenders toward AI-driven tools that can match that pace. By building a product around multiple AI labs' models rather than a single proprietary one, Palo Alto Networks is positioning itself less as a competitor to the model builders and more as the platform that decides which of their models gets deployed against a given customer's defense needs, a role that could prove durable even as the underlying AI models themselves keep changing.