Torvalds Tightens Linux Kernel Rules To Reject Deluge Of Low-Value AI Fixes

For about the past week, Linux kernel creator Linus Torvalds has been voicing complaints about "the continued flood of AI reports" making the list of security fixes unmanageable due to duplicate entries, and says it is now starting to choke the flow of new fixes in general due to the size of the Linux 7.1 Release Candidate 5 (rc5) becoming much larger than it should be. It's not that Torvalds has become strictly anti-AI, but rather that it's only good when it's actually helps.

Last week, the problem was that AI tools were producing duplicate bug reports from different people, sometimes even finding issues that were already fixed. Even with AI working as intended, reckless use of it without due diligence results in lots of extra work for Torvalds and other people in charge of actually maintaining the kernel and picking what fixes or additions get added with each update.

This week, AI isn't the only culprit behind rc5 becoming so unreasonably large. The bulk of it is "trivial stuff" fixing or adding random drivers, according to Torvalds, which isn't traditionally the purpose of release candidate updates. Usually, release candidate updates are meant to address critical fixes, including security problems like SSH-keysign-pwn or Copy Fail and actual regressions in functionality. While AI can be (and is) blamed for a lot of minor fixes being stuffed into the release candidate window, the other problem is with the developers themselves focusing the tools on non-essential fixes.



Image Credit: Reto Scheiwiller on Pixabay,