The risk of that is relatively low for kernel contributions, though. Most of the work being done is porting existing protocols/firmware into the latest Linux kernel, not creating novel features.

The larger risk is instability caused by *bad, hallucinated* code because it was submitted under the assumption of human authorship. In both cases, further review by the Linux team can be done if they understand where that code is coming from.

Banning AI does nothing, because theres no way of knowing who uses it without proper disclosure, which wouldnt happen if it were banned. To use an example from the article, it would be like banning code written with the use of a specific brand of keyboard.

Better to have it properly disclosed than to make it illicit