New Delhi: OpenAI wants to “Patch the Planet” but not in the way that environmentalists would hope. The AI giant on Monday announced a series of initiatives to deepen Daybreak, a project trying to improve global cybersecurity.
“We’ve applied our models to discover and generate patches for critical vulnerabilities in major browsers, network infrastructure, and operating systems such as FreeBSD and the Linux kernel,” read the OpenAI statement.
At the heart of this announcement is a simple fact—every day people rely on invisible open-source software that keeps a vast majority of websites and devices running. These are software whose code is freely available for anyone to inspect or modify, and yet it is maintained by a handful of people, mostly volunteers.
Now that AI is capable of identifying weaknesses in a software, it could potentially find bugs at a high speed. AI can scan millions of lines of code a lot faster than any human. But the dilemma comes after that—who will go through the bug reports, some of which are inaccurate, and then fix them?
This is where Patch the Planet steps in. A ‘patch’ is essentially a software update to repair the weakness that the AI has detected. In its recent announcement, OpenAI stated that it has partnered with Trail of Bits, a leading cybersecurity research, engineering, and consulting firm, supporting engineers to assess security risks, validate bug reports, and even write the ‘patches’.
“Patch the Planet is an internet-scale effort to help open source software get ahead of AI bug hunting tools,” Trail of Bits CEO and cofounder Dan Guido told WIRED. “But it’s also an effort to help the open source community see the benefits and not just the downsides of AI coding tools.”
OpenAI’s recent announcement also said that it has launched the full version of GPT-5.5-Cyber, an AI model built to focus on cybersecurity. This comes at a time when the company’s biggest competitor, Anthropic, had to pull out its new Fable 5 and Mythos 5 models off the market because of fear regarding what the models could be capable of.