Palo Alto Networks launches AI-powered defense using Claude Mythos and GPT-5.6-Cyber models

Palo Alto Networks Deploys Frontier AI Models for Always-On Security Testing

Palo Alto Networks announced Unit 42 Continuous Frontier AI Defense, an agentic offensive security service that leverages Anthropic Claude Mythos 5 and OpenAI GPT-5.6-Cyber to continuously test for vulnerabilities across enterprise systems1

1

2

2

. The service marks a strategic response to threat actors deploying AI to compress breach cycles by almost 97% in some cases, reducing attack timelines from weeks to hours4

4

5

5

. This development underscores how AI cybersecurity has evolved from a defensive advantage to an operational necessity as attackers gain machine-speed capabilities.

Multi-Model Approach Addresses Coverage Gaps in Enterprise Security Exposures

Palo Alto Networks' internal testing revealed that no single AI model caught more than 40% of vulnerabilities in complex environments, while Anthropic Claude Mythos and OpenAI GPT-5.6-Cyber found overlapping vulnerabilities less than 10% of the time3

3

. Unit 42 addresses this gap through a proprietary multi-model harness that routes security tasks to the optimal frontier AI models, improving efficacy while managing costs at scale2

2

4

4

. The service combines these gated capability models with open-weight models and Unit 42 threat intelligence to deliver comprehensive threat exposure management5

5

.

Continuous Testing Engine Validates Real-World Attack Paths

The Continuous Frontier AI Defense service provides a full-estate baseline scan followed by always-on testing as organizations' environments change, targeting web applications, APIs, cloud infrastructure, source code repositories, and network assets1

1

2

2

. Beyond identifying flaws, the service proves real-world exploitability by validating end-to-end attack paths and attempting to demonstrate how individual vulnerabilities can be chained together3

3

5

5

. This approach builds on Unit 42's Frontier AI Defense launched in April, which offered point-in-time exposure analysis4

4

.

Accelerated Remediation Through Code-Level Fixes and Virtual Patching

The service delivers actionable fixes with code-level guidance and virtual patch recommendations, enabling organizations to mitigate vulnerabilities before official patches exist1

1

5

5

. Customers can pair Continuous Frontier AI Defense with virtual patching tools to address security gaps immediately3

3

. Sam Rubin, Senior Vice President of Unit 42 at Palo Alto Networks, stated: "AI has created an asymmetric advantage for threat actors against organizations trying to defend at human speed. Modern cybersecurity requires machine-speed defense"2

2

4

4

.

Proven Results Show Year's Worth of Exposures Found in Three Weeks

Palo Alto Networks invested $17 million in R&D and validated the approach across more than 100 Unit 42 customer engagements over six months5

5

. During internal deployment, the company discovered a year's worth of exposures in three weeks2

2

5

5

. Customer assessments using the Frontier AI Exposure Analysis found exposures in 100% of tested organizations, with 37% rated high or critical severity4

4

5

5

. Most exposures stemmed from first-party apps, and more than two in three exposures in third-party apps had no known CVE5

5

.

Gated Capability Models Deliver Advanced Cyber Capabilities

Michael Moore, Cybersecurity Lead at Anthropic, noted that "Claude Mythos found flaws that survived decades of human review, and more than ten thousand high-severity vulnerabilities across the software the world runs on"2

2

4

4

. Access to Anthropic Claude Mythos has been tightly controlled, with Anthropic briefly suspending access in June to comply with US export controls before restoring it on July 12

2

. McCall McIntyre, Head of Global Cyber Partnerships at OpenAI, emphasized that "frontier AI is compressing the time it takes to find and exploit vulnerabilities, and defenders need frontier AI capabilities within the tools, workflows, and services they already trust"5

5

.

Global Availability Through Annual Subscriptions with Flexible Pricing

Continuous Frontier AI Defense is available worldwide through annual subscriptions, with pricing varying based on the specific mix of OpenAI, Anthropic, and open-source models customers select1

1

2

2

5

5

. All subscriptions leverage the multi-model harness to match optimal AI models to each security task4

4

. Palo Alto Networks CEO Nikesh Arora told Axios that tapping multiple AI models alongside offensive security expertise represents the industry's future, especially as each model trains on different data3

3

. Arora added: "If you believe that half of what's going to happen or be possible with AI hasn't been invented yet, then half of that security is missing, too"3

3

.