The creator of ChatGPT aims to reduce the security risks associated with open-source projects.

      OpenAI has introduced Patch the Planet, a new initiative designed to address one of the internet's overlooked issues – the persistent lack of funding for the security of open-source software.

      Patch the Planet combines OpenAI’s most security-focused AI models with Trail of Bits, a security company that has devoted its entire research team to this initiative, with additional support from HackerOne and Calif.

      How OpenAI intends to address the bugs

      The issue OpenAI aims to tackle is both real and specific. AI tools are capable of generating numerous potential vulnerabilities, yet overloaded maintainers must differentiate between actual threats and false alarms.

      Fouad Matin, OpenAI’s cyber tech lead, stated that maintainers engage in this work out of their passion for open source but are now overwhelmed by low-quality, AI-generated bug reports. Dan Guido, CEO of Trail of Bits, supported this viewpoint, describing the project as a significant endeavor to help open-source software stay ahead of AI bug detection tools while also highlighting the advantages of AI coding tools, not just their drawbacks.

      Researchers utilize OpenAI’s Codex Security and GPT-5.5-Cyber models to examine and validate issues, and every finding is personally reviewed before being presented to a maintainer. OpenAI is also subsidizing approximately 20 trillion tokens worth of Codex Security usage for both open-source and private code.

      The significance beyond bug fixes

      Over 30 projects are already involved, including cURL, Python, and the Go project, as Trail of Bits conducts an initial sprint using 20% of its workforce. In just the first week, this initiative has already identified hundreds of bugs and implemented numerous patches.

      This announcement comes at a time when competitor Anthropic faced scrutiny and had to withdraw its Mythos 5 and Fable 5 models from the market due to White House concerns regarding AI cybersecurity capacities. Reports indicate that OpenAI’s updated GPT-5.5-Cyber scores higher than Mythos 5 on the CyberGym benchmark, with 85.6% compared to 83.8%.

      Although that benchmark difference may seem minor, it serves as a reminder that the competition among AI labs could significantly influence internet security beyond any single product release.