Exactly. Even if an LLM wrote part (or all) of the code, we are still 100% responsible for what we commit and merge.
If I make a typo due to auto-complete, that's still on me for not properly re-reading the message before I hit send.
If I copy/paste a code snippet from Stack Overflow (R.I.P.) and it doesn't work, it's still my responsibility to debug that before it's done.
If I'm in a "self driving" car, it's still my responsibility to make sure I don't crash.
"The AI wrote it" is a strange defense for a bug. I still want to walk through the failure path. We own what happens when that code runs.
commit 0e0c119