- OpenAI is releasing a more permissive version of its cybersecurity model, which the company says is designed for advanced, authorized security work.
- The updated model, GPT-5.5-Cyber, is only available to vetted cybersecurity companies and researchers and is described as "more permissive and more capable for advanced, authorized cybersecurity work," according to a blog post.
- The updated model can perform deeper analysis across large codebases, identify security-relevant components, validate likely vulnerabilities, and develop and test software patches.
- OpenAI says the updated GPT-5.5-Cyber achieved an 85.6% score on CyberGym, an internal benchmark that measures whether an AI agent can reproduce known software vulnerabilities in testing environments, compared with 81.8% for GPT-5.5.
- OpenAI is rolling out a series of new programs and capabilities designed to let vetted cybersecurity companies use its models to help secure customer environments.
- The company is launching the OpenAI Daybreak Cyber Partner Program, which allows participating security vendors to use GPT-5.5 with Trusted Access for Cyber in the products and services they provide to customers.
- OpenAI is also helping fund Patch the Planet, an initiative founded with Trail of Bits and developed in collaboration with HackerOne and Calif, aimed at managing and remediating vulnerabilities identified with AI-assisted tools.
- OpenAI says it has established partnerships with Australia, Canada, France, Germany, Japan, Poland, the Republic of Korea and EU institutions.
OpenAI has unveiled its latest cybersecurity model, GPT-5.5-Cyber, aimed at enhancing security measures for authorized users. This model is designed specifically for vetted cybersecurity companies and researchers, offering advanced capabilities for security tasks.124578
The updated model can perform deeper analysis across large codebases, identify security-relevant components, validate likely vulnerabilities, and develop and test software patches. OpenAI claims that GPT-5.5-Cyber achieved an 85.6% score on CyberGym, surpassing the previous version's score of 81.8%.3
In addition to the model's release, OpenAI is launching the OpenAI Daybreak Cyber Partner Program, which enables participating security vendors to integrate GPT-5.5 with Trusted Access for Cyber into their products and services. This initiative aims to bolster cybersecurity efforts across various sectors.6
Furthermore, OpenAI is collaborating with organizations like Patch the Planet, founded with Trail of Bits, to manage and remediate vulnerabilities identified through AI-assisted tools. The company has also established partnerships with several countries, including Australia, Canada, France, Germany, Japan, Poland, the Republic of Korea, and various EU institutions, to enhance global cybersecurity efforts.
“OpenAI is releasing a more capable version of its GPT-5.5-Cyber model, designed for advanced cybersecurity tasks. The updated model aims to enhance security analysis and vulnerability management for vetted companies.”
