This is the bug that taught me not to trust AI with catching it by itself.

I found out that while solving the bug, I could introduce another one elsewhere. The original feature was fine, but an existing feature was somehow affected.

From now on, after each big change, I check if the original feature is OK, if the other related features are working as expected, and if the input data is being checked correctly.

It takes a little bit of time, but it is far better than finding it out from the user.

What do you do when building with AI to ensure that you won't create a new issue while fixing one?