The announcement that NetBSD is banning AI generated code is a noble gesture and all, but the main problem, as always, is detection. If everyone is acting in good faith, then all is well. But if someone generated code using Copilot then it may not be easily distinguishable from hand-written code. It's not like code has a watermark or anything like that. It's text.
I fear that this might end up with people getting falsely accused of submitting AI code. How does one prove their innocence?
Add comment