OpenAI Launches Patch the Planet to Fix Open-Source Bugs with AI

OpenAI's new Patch the Planet initiative pairs AI with security researchers to find and fix vulnerabilities in major open-source projects.

AI Tutorials · · 2 min read

Quick answer

OpenAI launched Patch the Planet on June 22, 2026 — a new initiative that pairs its GPT-5.5-Cyber model with security researchers to find and fix vulnerabilities in widely used open-source software. More than 30 projects including Python, cURL, and Go have signed on, and the fully released GPT-5.5-Cyber now scores 85.6% on the CyberGym benchmark.

OpenAI announced Patch the Planet on June 22 — a new initiative that uses AI to find and fix security vulnerabilities in widely used open-source software. The programme launched alongside the full release of GPT-5.5-Cyber, OpenAI’s most capable cybersecurity model.

What Is Patch the Planet?

Most of the software that runs the internet depends on open-source projects maintained by small teams with limited resources. Security bugs in these projects can affect millions of users, but maintainers often lack the time or tooling to find and fix them all.

Patch the Planet aims to close that gap. Co-founded with security firm Trail of Bits and backed by HackerOne, the initiative funds expert security researchers and equips them with OpenAI’s Codex Security tools and GPT-5.5-Cyber to work directly with open-source maintainers. More than 30 projects have signed on, including Python, cURL, the Go project, Sigstore, and pyca/cryptography.

GPT-5.5-Cyber Goes Full Release

The GPT-5.5-Cyber model, previously available only in limited preview, is now fully released to vetted defenders through OpenAI’s Trusted Access for Cyber programme. The updated model scores 85.6% on the CyberGym benchmark — up from 81.8% for standard GPT-5.5 — and can trace attack paths across large codebases, validate whether vulnerabilities are exploitable, generate targeted patches, and prepare evidence for human review.

OpenAI says the motivation behind this shift is practical: its models now find vulnerabilities faster than human teams can fix them, creating a bottleneck at the patching stage. Patch the Planet is designed to address that by putting AI-assisted remediation directly in the hands of maintainers.

What This Means for You

If you use any software — and you do — this matters. Open-source libraries power everything from web browsers to banking apps. Faster patching of security bugs in projects like Python and cURL makes the tools you rely on every day more secure.

For developers, the combination of AI-powered vulnerability detection and automated patch generation could meaningfully reduce the time spent on security maintenance. If you work with open-source code, keep an eye on whether your dependencies join the programme.

To learn more about how AI is changing software development, check out our guide to coding with AI or explore our Claude tutorial to see how AI assistants handle technical workflows. Sign up for our newsletter to get stories like this delivered daily.

Frequently asked questions

What is OpenAI's Patch the Planet initiative?
Patch the Planet is an open-source security initiative launched under OpenAI's Daybreak programme. It funds expert researchers and equips them with AI tools to find and fix vulnerabilities in widely used open-source projects like Python, cURL, and Go.
What is GPT-5.5-Cyber and who can use it?
GPT-5.5-Cyber is a specialised version of GPT-5.5 built for cybersecurity professionals. It can navigate large codebases, identify vulnerabilities, generate patches, and validate fixes. Access is restricted to vetted defenders through OpenAI's Trusted Access for Cyber programme.
How does GPT-5.5-Cyber perform compared to GPT-5.5?
GPT-5.5-Cyber scores 85.6% on the CyberGym benchmark compared to 81.8% for the standard GPT-5.5. It also scores 39.5% on ExploitGym versus 25.95% for the base model, showing significantly stronger security-specific capabilities.
Which open-source projects are part of Patch the Planet?
More than 30 projects have committed to participating, including cURL, the Go project, Python, Sigstore, and pyca/cryptography. OpenAI co-founded the initiative with security firm Trail of Bits in collaboration with HackerOne.

Want to keep learning?

Explore our guided learning paths or try building something with AI right now.

Enjoyed this article?

Subscribe for more AI insights delivered to your inbox every week.

No spam. Unsubscribe anytime.