It may not hallucinate yours to zero, or mine, or franks, or Mary's, but at some point it will do it to someone. That's the issue I have with these approaches *at scale*.
I'm sure one day we'll get 100.00% reliable outputs such that autonomous agents can do this, however that's not today.
I'm sure one day we'll get 100.00% reliable outputs such that autonomous agents can do this, however that's not today.