The risk of AI cheating in mathematical proof checks

New Scientist reports that formalisation lets computers check the logic of mathematical theorems and expose flaws. Developers of these tools have realised AI models could cheat their way to success, the outlet says, and are working to tighten the tools.