Putting OpenAI Cyber Models to Work for Defenders
Palo Alto Networks' Unit 42 integrates OpenAI's frontier cyber models into customer environments to identify and prioritize exploitable attack paths, including those without known CVEs.
Unit 42 has deployed frontier AI models, including OpenAI’s GPT-5.6 Daybreak, to analyze customer systems for vulnerabilities, testing exploitability and validating attack paths under expert supervision. The service identifies exposures that traditional scanners miss, with 36% of findings not linked to known CVEs due to complex, multi-gap weaknesses. Palo Alto Networks emphasizes a multi-model approach, routing tasks to the most effective AI system to balance performance and cost while adapting to new models as they emerge. The Frontier AI Exposure Analysis service combines AI-driven insights with Unit 42’s offensive security expertise and threat intelligence to help defenders prioritize fixes based on real-world attack potential.
The integration allows defenders to assess risks from an attacker’s perspective, chaining vulnerabilities across applications, infrastructure, identity, and cloud environments to determine which paths could lead to compromise. This adversarial approach compresses attack cycle timelines, enabling rapid identification of critical exposures before they can be exploited. Unit 42’s early deployments have demonstrated the practical value of AI in reducing false positives and focusing remediation efforts on high-impact vulnerabilities. The service is designed to address the overload of security findings many teams face, shifting focus from detection to actionable prioritization.
Palo Alto Networks highlights that frontier AI capabilities, previously limited to a few organizations, are now being applied directly in customer environments to strengthen defenses. The company notes that GPT-5.6 Daybreak, a model not previously available for commercial use, is now part of this offering. By leveraging these models, Unit 42 aims to stay ahead of AI-enabled threats, which the company warns are emerging faster than many organizations anticipate. The service builds on Unit 42’s existing Frontier AI Defense framework, which has already been adopted by hundreds of customers worldwide.
The announcement underscores the urgency for defenders to adopt advanced AI tools as attackers increasingly leverage similar technologies. Palo Alto Networks positions its Frontier AI Defense as a proactive measure, integrating AI models directly into security workflows to validate exposures and guide remediation. The company advises organizations to act quickly, as the window to implement such defenses is closing. More details are available on Palo Alto Networks’ Frontier AI Defense page, with additional resources provided upon request.