- Sources: report, Reddit thread
- Summary: Greg Kroah-Hartman announced that drivers/staging will reject LLM-generated patches, with tested security fixes as the stated exception. The reason he gives is purpose rather than code quality: staging exists so new developers learn on low-hanging fruit. He also states that with current and next-generation tools at least one third of the results they generate are flat out wrong or harmful, which is the stated reason the security-fix exception requires testing on actual hardware with the testing described. The announcement is carried here as Phoronix reproduces it, because lore.kernel.org remains behind the Anubis proof-of-work gate and did not resolve from this run.
- Why it matters: The exception is written as an evidence burden rather than a ban, which moves the cost of an LLM-assisted patch onto the sender in hardware time instead of onto the maintainer in review time.
- Follow-up: Confirm the wording against the kernel mailing list archive once lore.kernel.org resolves, and watch whether other subsystems adopt the same rule.
send feedback on this story