Key Takeaways:
- Palo Alto Networks launched a new artificial intelligence security testing service.
- The platform uses advanced AI models to scan company systems for security weaknesses.
- Company executives say automated tools help organizations protect their networks faster.
Palo Alto Networks launched Unit 42 Continuous Frontier AI Defense in Santa Clara on September 22, 2026. The new Palo Alto Networks AI service continuously tests company systems to help businesses find and fix security weaknesses quickly.
Service Details And New Capabilities
The Palo Alto Networks introduced a new global subscription service designed to protect company systems from advanced cyber threats. The platform uses AI models from Anthropic and OpenAI to continuously scan company networks.
The Palo Alto Networks AI system includes a continuous testing tool that runs regular checks across digital environments. It also uses its own multi-model system to send each task to the model that is best suited for it.
The system uses selected models, including Anthropic Claude Mythos 5 and OpenAI GPT-5.6-Cyber. These models also work with open-weight models, which gives the service access to a wider range of tools for security testing.
The service copies real-world attacks across web applications, application programming interfaces, cloud systems, and code repositories. It then suggests which problems should be fixed first and provides virtual patch recommendations to help security teams stop threats.
Organizations can also use the service with Frontier Virtual Patching to add protection before a vulnerability is publicly disclosed. This can help companies defend their networks against new attacks before they become widely known.
Internal Testing And Customer Results
Palo Alto Networks spent six months testing the new service internally before launching it publicly. During those tests, the system found a year’s worth of security weaknesses in just three weeks of operation.
The company invested $17 million in research and development for the project. Its engineers used the funding to improve the testing methods and the platform over several months.
Tests with customers found security weaknesses in every organization included in the evaluation. Company officials said 37 percent of those weaknesses were rated high or critical in severity.
Most of the vulnerabilities were found in applications built by the organizations themselves. More than two out of three weaknesses found in third-party applications did not have a known vulnerability identifier.
Executives Discuss Faster Cyber Defense
Sam Rubin said automated tools give security teams an important advantage when dealing with attackers.
McCall McIntyre of OpenAI said advanced AI models are reducing the time needed to find software flaws. Michael Moore of Anthropic added that these models can identify thousands of serious vulnerabilities across software systems.
Palo Alto Networks plans to offer the Palo Alto Networks AI service worldwide through an annual subscription for corporate customers. Organizations will be able to choose different model setups based on their security needs and budgets.
The new service builds on other AI security tools the company released earlier this year.
Trusted by more than 75,000 customers worldwide, Palo Alto Networks provides cybersecurity solutions across networks. The new service combines threat information with AI-based systems to support security testing.
