A result that generalizes prompt hacking and would point towards most deep learning AIs are exploitable (and proably trivially so).
A result that generalizes prompt hacking and would point towards most deep learning AIs are exploitable (and proably trivially so).