Palo Alto Networks launches Unit 42 Continuous Frontier AI Defense

Palo Alto Networks has launched Unit 42 Continuous Frontier AI Defense, an annual subscription service designed to provide ongoing offensive security testing. The service utilizes a multi-model approach, combining gated frontier models such as Anthropic’s Claude Mythos 5 and OpenAI’s GPT-5. 6-Cyber with various open-weight models to identify and validate vulnerabilities.

The service aims to protect web applications, APIs, cloud infrastructure, code repositories, and network assets. By using a proprietary harness to route tasks to the most suitable model, the platform provides prioritized fixes, code-level guidance, and virtual patch recommendations. This continuous testing model addresses the reality that no single AI model can identify all vulnerabilities, as research indicates individual models may catch less than 40% of exposures.

This launch is part of Palo Alto Networks’ broader strategy to integrate AI security into its platform, following recent acquisitions such as Koi and Console. The company is positioning itself to capture recurring revenue from the rapid adoption of AI by both defenders and attackers.