Palo Alto unit probes systems with Anthropic and OpenAI models
Palo Alto Networks' Unit 42 has launched Continuous Frontier AI Defense, a subscription service that uses access-restricted models — Anthropic's Claude Mythos 5, OpenAI's GPT-5.6-Cyber, and open-weight models — to continuously test customers' web apps, APIs, cloud infrastructure, code repositories and networks. Unit 42 assigns models per task, attempts to prove exploitability, maps likely attacker paths, and suggests fixes or virtual patches. Palo Alto says development took six months and $17 million across more than 100 engagements, and found exposures at every tested customer, with 37% rated high or critical severity.
Palo Alto offers continuous AI-driven scanning using Claude Mythos and GPT-5.6-Cyber
Context
Attackers began using AI to find and exploit flaws far faster. Palo Alto responded by building around-the-clock AI testing. This could lead to more automated defenses and more frequent vulnerability discovery.
The full analysis
19 dimensions on this story — world impact, market read, and what happens next.
- Full ContextLocked
- Affected SectorsLocked
- Stock ImpactLocked
- Economic IndicatorLocked
- Investor RelevanceLocked
- Professional RelevanceLocked
- Watch PointsLocked
- Probability of ChangeLocked
- Debate PointsLocked
- Prerequisite KnowledgeLocked
- Follow-up QuestionsLocked
- Pros & ConsLocked