Frontier AI for Defenders: CrowdStrike and OpenAI TAC

CrowdStrike has been selected for OpenAI's Trusted Access for Cyber (TAC) program, as OpenAI released GPT-5.4-Cyber, a model designed for defensive cybersecurity. This collaboration enhances CrowdStrike's role in secure AI adoption, providing verified defenders governed access through identity verification and tiered controls. The integration of OpenAI models within the CrowdStrike Falcon platform is complemented by AgentWorks, a framework allowing the selection of optimal models for specific security tasks while maintaining enterprise-level governance. The 2026 Global Threat Report by CrowdStrike highlights the rapid automation of attacks, with the fastest eCrime breakout time recorded at 27 seconds in 2025. This underscores the importance of discerning exploitable vulnerabilities within enterprises, leveraging data from over 280 tracked adversary groups. With visibility into over 1,800 AI applications, CrowdStrike addresses governance challenges by monitoring AI agent activities at endpoints. As the EU AI Act's next phase approaches, the need for robust technical infrastructure becomes imperative, with CrowdStrike positioned to deliver comprehensive intelligence, protection, and governance.