Palo Alto Networks Launches Unit 42 Continuous Frontier AI Defense

Palo Alto Networks announced Unit 42 Continuous Frontier AI Defense on 22 September 2026: a subscription service that continuously tests an organization’s systems for security exposures using frontier AI models (Palo Alto Networks press release).

What the service does

According to the announcement, the service runs a continuous testing engine against a customer’s infrastructure, simulates attackers to validate full attack paths, and returns prioritized fixes with code-level guidance. It routes tasks across several AI models, which the company names as Anthropic’s Claude Mythos 5, OpenAI’s GPT-5.6-Cyber and open-weight models. It is sold worldwide as an annual subscription.

The claims, and how to read them

Palo Alto says its internal testing found “a year’s worth of exposures in three weeks”, that its customer assessments found exposures in every customer tested with 37% rated high or critical, and that attackers using AI have compressed breach cycles “by almost 97% in some cases”. These are the vendor’s own figures from a product launch, not independent measurements.

What it means if you’re not a large enterprise

The underlying point holds at any size: attackers scan continuously, so checking your exposure once a quarter leaves long gaps. You can cover the basics for free: run a port scan on your public IPs, check your SSL certificates and security headers, and repeat after every infrastructure change.