On Tue, May 19, 2026 at 3:03 PM Danilo Krummrich <[email protected]> wrote: > > That said, I expect LLMs to be (heavily) used to write drivers, and LLMs > are typically trained to be compliant. So, I could imagine that a prompt > along the lines of "Can't we make this work without unsafe somehow?" > leads > to an LLM abusing this, while going unnoticed for the user.
Could we steer LLMs to avoid that? e.g. adding a line about `Unsafe*` items to Sashiko's prompts. Also, a `checkpatch.pl` (or similar) warning may be doable too, since false positives are OK there. It could go to the ASH (Acknowledged Soundness Holes) list we discussed some times if we end up adding it. Cheers, Miguel
