Palo Alto Launches Continuous AI Cybersecurity Testing Service

Palo Alto Networks launched a cybersecurity service on September 22 that uses Anthropic’s Claude Mythos 5, OpenAI’s GPT-5.6-Cyber and open-weight models to find weaknesses in corporate systems. The company says its Unit 42 service tests continuously and helps customers prioritize fixes.

The service checks web applications, APIs, cloud infrastructure, code repositories and network assets as a customer’s environment changes. It also links separate flaws into possible attack paths, helping teams judge whether a weakness can be exploited in practice.

Palo Alto says its tests found that no single AI model detected more than 40% of vulnerabilities in complex environments. The company also reported less than 10% overlap between the vulnerabilities found by Claude Mythos 5 and GPT-5.6-Cyber, a gap it says supports using multiple models.

The company developed and tested the approach over six months and more than 100 customer engagements, according to its announcement. In its own systems, Palo Alto said the testing found a year’s worth of exposures in three weeks. In customer assessments, 37% of identified exposures were rated high or critical.

Unit 42’s output includes prioritized recommendations, code-level guidance and virtual patches. Those temporary fixes can help customers reduce risk while waiting for an official software update, Palo Alto said.

The service is available worldwide through annual subscriptions. Pricing varies with the mix of Anthropic, OpenAI and open-source models selected, and the company did not disclose subscription prices.

The launch builds on Unit 42’s earlier frontier-AI security offering, introduced in April. For customers, a key consideration will be whether continuous scanning and faster remediation justify the cost of using restricted, advanced models alongside security specialists.