1 paper touches this idea.
Papers
A hidden AI bias you can't see in its text? Force it to leak.
Related concepts
Ideas that show up alongside Bias Detection In Language Models.