Palo Alto Networks launches always-on AI security testing on Claude Mythos and GPT-5.6-Cyber

Palo Alto Networks' Unit 42 unveiled Continuous Frontier AI Defense, a subscription that uses Anthropic's Claude Mythos 5, OpenAI's GPT-5.6-Cyber and open-weight models to hunt for weaknesses in customers' web apps, APIs, cloud, code and networks around the clock, then prove exploitability and suggest fixes. The company spent six months and $17m on the approach across 100+ engagements; testing found exposures at every customer, 37% rated high or critical.

• Service runs on Claude Mythos 5, GPT-5.6-Cyber and open-weight models
• Built over 6 months and $17m across 100+ customer engagements
• Exposures found at every customer tested, 37% high or critical
• Anthropic: Mythos surfaced over 10,000 high-severity vulnerabilities