OpenAI launches Patch the Planet to help fix open source security bugs

In short: OpenAI announced a new program called Patch the Planet to help open source software teams find and fix security problems, and it also shared updates to its cyber-focused AI tools.

What happened

OpenAI said it is starting Patch the Planet with security firm Trail of Bits, and working with HackerOne and Calif. The goal is to help maintainers of open source software, which is code that anyone can use and often runs behind the scenes in other products.

OpenAI and its partners say open source maintainers are often volunteers with limited time. They can get overwhelmed by bug reports, especially as AI tools make it easier to produce large amounts of low quality reports. OpenAI’s cyber tech lead Fouad Matin said maintainers are spending time reviewing “slop CVEs,” meaning messy or unhelpful vulnerability reports.

The program offers free security consulting to participating projects. That includes checking a project’s codebase (its main body of code), validating whether reported issues are real, and helping create and submit fixes, also called patches (like repairing a hole in a roof before the next storm).

OpenAI also announced an improved version of its limited access cyber model, GPT-5.5-Cyber, under its “Trusted Access for Cyber” program, which is not a public release. OpenAI said the model scores 85.6% on CyberGym, a test used to compare cyber capabilities.

Why it matters

Open source software is part of many everyday services, even when people never see it directly. If widely used open source projects fall behind on security fixes, that can create real risks for companies, governments, and users. Patch the Planet is one attempt to reduce the workload on the small teams responsible for maintaining that shared infrastructure.

Source: Wired