> I’m willing to believe that the LLM-powered Sashiko code reviews provide a lot of good insights. However, to address just one externality of this tool, consider that AI companies are driving up the price of consumer hardware. An AI powered code review may improve a patch, but that patch won’t be of much use to the increasingly large cohort of people who are being priced out of the hardware they could run Linux on to enjoy the better patch.

The Linux kernel also owes its users the highest degree of security possible. By proposing a hardline stance against valid security reviews generated by AI, the author is ignoring some other very real, possibly even fatal externalities just for the sake of their political purity.