Depends on the exact scenario.

We’re also dealing with language differences. English is not the developer’s first language. What may seem a clear sentence to a native speaker, could be easily misinterpreted/mistranslated to something similar, but different enough that the answer changes.

It seems that the AI use was early in development, and limited to temporary placeholders that were going to be replaced. Since they were patched out within days of release, that seems to imply they already had replacement assets on hand, they were just missed during final checks before release.

The answer from the devs also changed prior to the awards show that implies that they may have had an updated interpretation of the qualification question or answer. If they thought the question was about AI use in the final product, then accidentally missing a placeholder swap shouldn’t be disqualifying. Likewise, early experimentation with the tech and then deciding not to use it probably should not disqualify either. But if the qualification is a hard yes/no with absolutely no context or consideration whatsoever, then that’s a different outcome, and hence them clarifying for the awards team.

Personally I think the hard limit without any room for consideration or interpretation is a shit qualification. Especially considering that isn’t really the case for most awards. Look at the definition of “indie” for example. There’s a half dozen different interpretations people have ranging from having to be self published, avoiding just large publishers, or just the publisher not having creative influence. That’s a lot of interpretation comparatively.