OpenAI announced the upgrade of its Daybreak cyber defense toolchain and the official release of the full GPT-5.5-Cyber model tailored for cybersecurity defense scenarios. In the CyberGym benchmark, which evaluates AI agents' ability to reproduce known vulnerabilities, GPT-5.5-Cyber achieved a single-model score of 85.6%, outperforming GPT-5.5's 81.8% and Anthropic Mythos 5's 83.8%. This demonstrates the significant advantage of domain-specific models optimized for cybersecurity tasks.
From Vulnerability Discovery to Automated Remediation: The Shifting Bottleneck
OpenAI noted that as AI dramatically improves vulnerability discovery efficiency, the core bottleneck in cybersecurity is shifting from 'finding vulnerabilities' to 'automatically fixing them.' To address this, the company upgraded the Codex Security plugin, enabling developers to automatically analyze vulnerabilities and generate fix patches within the Codex environment. Since its preview release in March, Codex Security has scanned over 30 million code commits and autonomously confirmed fixes for 500,000 security defects. The new version also supports CodeQL query integration and SARIF standard file export, further automating and standardizing security analysis.
Open-Source Security Initiative and Industry Collaboration
Additionally, OpenAI, in partnership with Trail of Bits and HackerOne, launched the 'Patch the Planet' open-source security project. It provides ChatGPT Pro subscriptions and API credits to over 30 mainstream open-source projects such as cURL and Go, with patch quality manually verified by security expert teams. Under the Daybreak partner program, security vendors including Palo Alto Networks and Wiz have already integrated relevant capabilities, marking an accelerated deployment of AI-driven cybersecurity defense systems.