OpenAI has launched the GPT-5.5-Cyber model alongside the "Trusted Access for Cyber" (TAC) framework, aimed at bolstering cybersecurity defenses. The GPT-5.5-Cyber model is now available in a limited preview for defenders of critical infrastructure, enhancing specialized cybersecurity workflows. The TAC framework ensures that advanced AI capabilities are accessible only to verified cybersecurity defenders, reducing model refusal responses during tasks like vulnerability identification and malware analysis. Starting June 1, 2026, users will need to enable advanced account security to access these capabilities. OpenAI is partnering with security firms such as Cisco, CrowdStrike, and Palo Alto Networks to enhance the security ecosystem's defense cycle, focusing on improving efficiency in vulnerability research, patching, and supply chain security.