An asset was rejected or sent to review and you're not sure why. The answer is always in the asset's rule results - this article shows you how to find it and what to do if you disagree.
Find out which rule caused it
Open the run from Moderation Runs (for automations, Automations).
Use the Rejected or Needs Review tab to find the asset, and select it.
In the sidebar, open the Details tab.
Scan the Rule Results: the rule that failed (or wasn't confident enough) is your answer. Each rule shows its outcome, a confidence percentage, and an explanation of what was found - for example, "Text detected."
For rejected assets, the rejection reason configured on the rule appears too.
For videos, click any timestamp in an explanation to jump to the exact moment in the player.
Compare with your policy
If the run used a policy, remember: an asset must pass every rule in the policy. It's common for an asset to pass nine rules and be rejected by the tenth. The rule filter (the All Rules dropdown) lets you view results one rule at a time to see each rule's behavior across the whole run.
Why "Needs review" happens
"Needs review" isn't a failure - it means the rule's confidence fell between "clearly fine" and "clearly not." How often this happens is controlled by each rule's Review threshold:
A High threshold sends more assets to review (more human control).
A Low threshold decides more automatically (less review work).
If the result seems wrong
Override it. Select the asset and Approve or Reject it manually. Your decision wins over the automatic result and is recorded in the asset's History tab.
Tune the rule. If the same rule keeps flagging content you consider fine, edit the rule: adjust its Review threshold or refine its configuration (for example, make an AI Content Detection question more specific). Then test with a small one-time run.
Check the asset. Occasionally the rule is right in a way that's easy to miss - a faint watermark, tiny text, a slightly off background. The explanation and confidence score usually point at what the system saw.
Existing results never change when you edit a rule - only future runs use the new settings.