Greg Kroah-Hartman, responsible for maintaining the stable and staging branches of the kernel Linux, announced that it would no longer accept patches prepared using AI tools into the drivers/staging section. It is noted that such automatically generated patches contradict the core idea of using the drivers/staging subsystem as a training ground for new kernel developers. Code in drivers/staging is initially classified as problematic (TAINT_CRAP), and maintainers intentionally do not perform automated cleanup to allow novice developers to develop the skill of independently resolving minor kernel issues.
An exception is made only for AI-generated patches that address real security issues. Such driver fixes must be tested on real hardware and accompanied by a test report. The author submitting the patch must be able to prove that the fix is correct. Manual verification is necessary, as Greg estimates that about a third of potential security fixes generated by AI models, even the most modern ones, are incorrect or cause problems.
Source: opennet.ru
