I don't care about either of these so long as there is a human-only escalation path for potential false positive cheat detection, or for a potential mishandling of a support issue.
In fact, it would be smart to mitigate needless handling of common issues so long as the AI is capable of handling the issues properly the vast majority of the time. It should lead to faster turnaround for support tickets that are more complex as the support reps would be freed up from the mundane support tickets. If AI is being used to free up resources that can be used elsewhere rather than simply replacing the support reps and laying them off, then that's a win (again, provided that the vast majority of issues handled by AI are properly managed).