Comment Comment by jameslk Could models mitigate this by answering questions incorrectly with random information instead of outright refusing to answer them? parent jameslk 11 months ago
