Cloudinary Moderation is an AI-powered review layer for your visual content. It checks your images and videos against rules you define - quality standards, brand guidelines, legal and safety requirements - and tells you which assets are ready to use, which should be rejected, and which need a human decision. AI does the first pass; your team always makes the final call.
What you can check
Moderation reviews the assets that already live in your Cloudinary Media Library, or new assets as they arrive. Typical checks include:
💎 Quality and composition - blurry or low-resolution images, bad crops, off-center products, incorrect image size.
🎨 Brand consistency - missing or misplaced logos, watermarks, background color, unwanted text or promotional overlays.
🛡️ Legal and safety - AI-generated content, images that already exist on the public web, adult or unsafe content.
🪄 Anything you can describe - with AI Content Detection rules, you describe in plain language what should or shouldn't appear, and Moderation checks for it.
Both images and videos can be moderated. Video moderation uses its own set of rules and is available on eligible plans.
The building blocks
You'll see four main concepts across the app:
Rules - a rule is a single check, such as "the background must be white".
Each rule has settings you control, including how strict it should be before deciding automatically.Policies - a policy is a group of rules you save together and reuse, so every run applies the same standard. For example, a "Product images" policy might combine Image Quality, White Background, and Watermark Detection.
Moderation runs - a run is one round of moderation: a set of assets checked against your rules or policies.
Automations - an automation is moderation that keeps running on its own, checking new assets as they're uploaded or changed.
Two ways to run moderation
One-time moderation - you pick the assets, pick the rules or policies, and run once. Best for reviewing an existing folder, or for trying Moderation for the first time.
Ongoing moderation - you set it up once, and it runs automatically whenever a trigger happens, such as new assets being uploaded to a folder. Best when content keeps arriving and every asset should be checked.
What you get back
After a run, each asset receives one of three results:
✅ Approved - the asset passed all the selected rules.
❌ Rejected - the asset clearly failed at least one rule.
🚩 Needs review - the result wasn't clear-cut, so a person should take a look.
You can review every result, see exactly which rule was triggered and why, and approve or reject any asset yourself - your manual decision always wins.
Results can also be saved to your assets as structured metadata, so other teams and systems can act on them.