This gets repeated all the time, but it’s total nonsense. The output of an LLM is fixed just as the output of a human is.