Better Codes - and how they can be worse if you're not careful

An exploration of how an AI tried to work around a security feature to handle a theorem that was hard to prove.