OpenAI Expands Daybreak Cybersecurity Initiative with New Automated Patching Tools

The GPT-5.5-Cyber model has been officially launched by OpenAI as part of its expanded Daybreak cybersecurity initiative. This release aims to transition defensive security from vulnerability discovery to automated patch generation. The company stated that the tool is designed to help organizations secure software at scale.

Alongside the model, OpenAI released an update to its Codex Security plugin. Since its research preview in March, the plugin has scanned more than 30 million commits across 30,000 codebases. Human reviewers marked 70,000 findings as fixed, while over 500,000 findings were resolved automatically. The updated plugin allows developers to run deep scans, trace attack paths, and generate codebase-specific patches for software.

Capabilities of the GPT-5.5-Cyber model

The GPT-5.5-Cyber model demonstrated improved performance on several cybersecurity benchmarks. On CyberGym, which evaluates vulnerability reproduction, the model achieved a score of 85.6%, compared to 81.8% for the standard GPT-5.5. Additionally, the model scored 39.5% on ExploitGym and 69.8% on SEC-bench Pro, outperforming previous versions in long-horizon vulnerability discovery.

Collaborative Security Initiatives

OpenAI also introduced the Patch the Planet initiative in collaboration with Trail of Bits, HackerOne, and Calif. This project helps open-source maintainers fix vulnerabilities in widely used software. More than 30 open-source projects, including cURL, Go, Python, and Sigstore, have committed to participate. Furthermore, the Daybreak Cyber Partner Program allows security providers to integrate these tools into their services.

Global Government Partnerships

The company has established partnerships with several international governments to protect critical infrastructure. These include agreements with Australia, Canada, France, Germany, Japan, South Korea, and the United Kingdom. OpenAI is also working with the United States government, including the Center for AI Standards and Innovation, to test the GPT-5.5-Cyber model before deployment.

Future Outlook for Automated Defense

The expansion of these tools represents a shift toward automated, machine-speed defense. By providing verified defenders with access to the updated model, the initiative aims to reduce the window of exposure between vulnerability discovery and remediation. Organizations can now integrate these automated workflows to secure their digital infrastructure.