Libreboot, the Coreboot downstream, has dropped the anti-AI policy it put in place last month to forbid contributions made with AI or LLMs, Phoronix reports. In the commit that deleted the project’s AGENTS.md file, developer Leah Rowe wrote that she was never going to enforce the ban, that exceptions would pile up because of upstream code, and that she would not promise what she could not reliably deliver. “i got too excited writing it. i believe everything in it. but it’s an unenforceable policy.” The hard ban is gone from that file.
That is not the end of the story. Libreboot’s site now has a page titled “AI Slop Reduction Policy,” and it is careful to say this is not an outright ban. The project says it will make efforts, where feasible, to reduce the influence of AI on the code it uses, including reverting or rewriting some patches, but it also says Libreboot is a small project, depends on upstreams such as Coreboot that accept AI-generated work, and cannot fight the whole world. I am staying with those two public statements: the unenforceable ban was deleted, and the current page describes a softer, more limited effort rather than a new hard rule.
My take
I respect someone who would rather drop a promise than become a hypocrite about it. Whether a “reduce slop where feasible” policy can be applied consistently is a different question, and it will show up in the patches, not in the manifesto. I have not been following Libreboot’s tree commit by commit, and I will not pretend I have.
God Bless, and Use the Source!