How does a dishonest verifier affect plausible deniability of the interactive Schnorr protocol?
A dishonest verifier does not need to sample the challenge randomly. What if he just computes it as the evaluation of a one-way function on let's say a string "Alice authenticated, it's true". Later it can publish the string and the description of the function. What does this say about the deniability of the transcript of the protocol? Guess that should be sufficient to figure out the details on your own.