Launching 3 July — what does your AI-generated code keep getting wrong?

you let AI write a chunk of your code now, it compiles, tests pass, you ship. But "tests pass" was never "safe in production" — and AI is very good at code that looks right and quietly isn't (security holes, compliance gaps, the maintainability debt you feel six months later). Quality Clouds Hub checks for exactly that, on any stack, with a free tier you can point at real code.

I'm curious: what's the thing AI-generated code keeps getting wrong for you? The bug that looked fine in review and bit you later. Genuinely want to hear it — it sharpens what we flag.