🏒 Lucian Marin DeepSeek V4 Flash found 10 bugs in Subreply code. I fixed those bugs then asked again. Another 10 bugs. Fixed. Then I asked Opus 4.6 and it found 40 bugs from which I fixed 30. Then I asked ChatGPT 5.6 Terra and another 20 bugs from which I fixed 10. I had the impression I write simple and correct code, but AI is insane at this point and getting even more powerful.

Zero Edge I mean if they never turn up how serious can they be? I am sure I have a million bugs in my code but a lot of them are almost impossible circumstances.