Realistically I think AI is creating security vulnerabilities that are huge, glaring problems far outpacing the vulnerabilities they are catching.

Remember Moltbook, for example?