The heuristics that would be used to "detect AI" here would be things that shouldn't be happening anyway, so false positives wouldn't matter.