I don’t think anyone has mentioned this yet, but the idea of AI hacking tools has also led to a lot more human effort in finding and fixing security issues. A while back there was an article about AI security tools and something about over 200 security vulnerabilities patched (perhaps in Chrome?). But in the detail of the article, it mentioned only 4 were found by AI. The rest were found by humans trying to get there before people with AI tools found them.