Fine Print

Automatic checks · Checked November 16, 2026

Cloudinary Moderation

Cloudinary is a platform that stores, changes, and delivers images and videos. Cloudinary Moderation checks those assets against rules for brand consistency, quality, and compliance. Some rules are built in, for example a rule that finds well-known logos. You write the others as a yes or no question about the image or the video.

Each rule can approve or reject an asset when the result is clear. If the result is between the two thresholds, the asset goes to a person for review.

GroupAutomatic checks
Where the work must be
Assets in Cloudinary

Cloudinary Moderation checks assets that are in the Cloudinary Media Library, or new assets when they arrive there. Ongoing moderation checks new assets automatically. A check starts on upload into selected folders, when a user adds a tag, or when structured metadata matches a condition. One-time moderation checks a set of existing assets that the user chooses, once.5

Can an agent run it?
Partly: its API reads saved results

Cloudinary Moderation can store its results as metadata on each asset. An agent can read those results through the Cloudinary API. When Cloudinary Moderation writes a result to the structured metadata of an asset, Cloudinary sends a webhook notification. An outside service can act on that notification.1

File types
Images and videos

Cloudinary Moderation works on images and videos. For a video, it checks the frames, the on-screen text, and the audio track.1

What the check gives back
Result, confidence, and explanation

Each asset gets one result: Approved, Rejected, or Needs review. For each rule, the result shows the outcome, a confidence percentage, and an explanation of the finding. For a video, an explanation can include a timestamp that opens the video at that moment. The product page states that Cloudinary Moderation can suggest fixes, or apply fixes automatically, with Cloudinary transformations when possible.12

Who decides
People make the final call

The AI makes the first check, and people make the final decision. Each rule can approve or reject an asset automatically when the score is clear. If the score is between the two thresholds, the asset goes to a person for review. A person can approve, reject, or flag any asset manually. The manual decision overrides the automatic result, and Cloudinary records it with the name of the person. A one-time run changes nothing in the Media Library by itself. The user reviews first and then acts. If a user saves the results as structured metadata, a workflow can act on them automatically. Examples are to make an approved asset public or to move a rejected asset to a quarantine folder.1

Where the rules come from
Built-in and custom rules

The rules are built-in checks and custom rules. A customer can start with pre-defined rules and add custom rules for its brand guidelines. With the Image Content Detection rule, the customer writes a yes or no question about the image in plain language. Each rule has a Review threshold. It sets how confident the check must be before it decides automatically. A policy is a set of rules that a team saves and reuses. A policy covers one asset type, images or videos. The product page states that a custom-trained AI model learns the visual guidelines of the brand.1

Brands in one accountNot published
Plan with the check
Free plan. Pro plan for ongoing checks

Cloudinary Moderation has a free plan. Ongoing moderation, AI-generated detection, and web-sourced image detection are available only on advanced plans. Pro rule types need a Pro plan. Ongoing moderation with automations is also a Pro feature. Video moderation uses its own set of rules. It is available on eligible plans.1

Free plan
Free plan with a usage limit

Cloudinary Moderation includes a free plan. A customer can use it to test moderation results on its own assets, up to a limit of moderation actions. The Cloudinary Image and Video APIs also have a Free plan at $0 with no credit card required. The DAM (digital asset management) product has a free trial for 30 days.1

What does Cloudinary check?

CheckResultWhat the vendor says
LogoWell-known logosThe Well-Known Logo Detection rule finds well-known brand logos in an image. A customer can use it to block logos or to require a logo, and can limit it to named brands. The Brand Element Placement rule checks the position, size, and spacing of well-known logos or promotional text in an image. The documentation lists unauthorized logos and missing logo spacing among its example checks for brand consistency.2
The vendor's words

What it checks: the presence of well-known brand logos in the image.

ColorBackground colorThe Background Color rule checks that the background of an image matches a required color. The color can be white or a custom hex value. The product page lists off-brand colors among the things that Cloudinary Moderation detects.2
The vendor's words

What it checks: that the background is clean, uncluttered, and matches the color you require

TypeNot publishedThe vendor's pages do not say.
Copy and termsVideo text and speechFor videos, the Video On-Screen Text Check rule checks the text on screen against a yes or no question that the customer writes. The Video Transcript Check rule does the same for the spoken words. For images, the Text Detection rule checks whether text appears in the image.4
The vendor's words

What it checks: the text that appears on screen (captions, overlays, signs, packaging), evaluated against a yes/no question you write.

Claims and legal linesVideo claims and disclaimersFor videos, Cloudinary names spoken claims, required disclaimers, and forbidden phrases as uses of the Video Transcript Check rule. The customer writes the yes or no question.4
The vendor's words

When to use it: spoken claims and compliance - required disclaimers, forbidden phrases, offensive language in voiceovers or interviews.

The fine print

Limits that Cloudinary states

  1. Longer videos, and video files over 500 MB, end with an error for that rule instead of a result.

    Rules that look at video frames check 600 frames for each video at most. That is about 10 minutes at 1 frame for each second. Longer videos and files over 500 MB end with an error.4

  2. AI-based checks are good, not perfect. Plan for human review where stakes are high

    Cloudinary states that AI-based checks are not perfect. It tells customers to plan for human review where the stakes are high.2

  3. Originality and logo matches can have lookalikes - verify before acting on a rejection.

    Logo matches and originality matches can be lookalikes. Cloudinary tells customers to verify a match before they act on a rejection.2

  4. a match means the image was found somewhere online - it doesn't say whether you have the rights to it.

    The Originality Check finds whether an image exists on the public web. A match does not say whether the customer has the rights to the image.2

  5. detection is probabilistic.

    AI-generated image detection is probabilistic. Borderline images go to Needs review, so that a person decides.2

  6. this is a technical metadata check - it reads the profile embedded in the file and does not analyze the visible colors in the image.

    The Color Profile rule reads the color profile that is embedded in the file. It does not analyze the visible colors in the image.2

  7. Automations moderate assets that arrive after they start, not earlier uploads.

    An automation moderates assets that arrive after it starts. It does not check earlier uploads.18

  8. Text that's on screen for a moment can fall between frames - raise the rate to catch it.

    On-screen text rules check a set number of frames for each second. Text that is on screen for a moment can fall between frames.18

  9. Coming Soon rule types can't be used yet.

    Some rule types are in Lab and some are marked Coming Soon. A customer must contact Cloudinary to try a Lab rule type, and cannot use a Coming Soon rule type yet.16

  10. A run checks one asset type at a time

    A run checks one asset type at a time, images or videos.13

  11. Existing results don't change when you edit a rule

    Existing results do not change when a customer edits a rule. Only later runs use the new settings.10

  12. Both one-time runs and automations draw from the same quota

    Each one-time run and each automation uses Moderation Actions from the quota of the plan.19

Which plans does Cloudinary have?

  • Cloudinary publishes no price for the advanced plans of Cloudinary Moderation. A customer must contact Cloudinary to discuss them.1
  • The pricing page gives plans for the Cloudinary Image and Video APIs, and no price for Cloudinary Moderation. The Free plan of the Image and Video APIs costs $0. It includes access to MCP (Model Context Protocol) servers.17
  • The Plus plan of the Cloudinary Image and Video APIs costs $99 for each month with monthly billing. With yearly billing, it costs $89 for each month. It adds paid add-ons and role-based administration for several users.17
  • The Advanced plan of the Cloudinary Image and Video APIs costs $249 for each month with monthly billing. With yearly billing, it costs $224 for each month. It adds custom domain support and authentication options.17
  • The Enterprise plan of the Cloudinary Image and Video APIs has no published price. A customer must contact Cloudinary. It adds a security and compliance review, enterprise support, and custom contracts.17

Prices are from the vendor's pricing page.

What else do Cloudinary's pages say?

Audit trail for each asset
The History tab of each asset shows automatic decisions and manual decisions, the person who made them, the time, and the reasons.20
Results are not saved by default
By default, a run writes nothing to the assets. A user can save the results as structured metadata, or can turn on automatic saving in Settings.7
AI-generated image detection
The AI-Generated Image Detection rule checks the provenance data in the image first. If the image has none, an AI detection model decides.2
Six areas of image rules
Image rules cover six areas: composition and background, custom checks with AI, library hygiene, originality and authenticity, quality and size, and text and logos.12
Five video rule types
Five video rule types are available: Video Content Detection, Video On-Screen Text Check, Video Transcript Check, Video Text Metadata Match, and Video Transcript Metadata Match.13
MCP servers manage assets
Cloudinary lists five MCP (Model Context Protocol) servers: Asset Management, Environment Config, Structured Metadata, Analysis, and MediaFlows. With them, an AI agent can upload, manage, transform, and analyze assets.21
Separate add-on moderation in the Upload API
The Cloudinary Upload API has a separate moderation parameter. It marks an asset for automatic moderation by an add-on, for example Amazon Rekognition AI Moderation or WebPurify, or for manual moderation.22
AI Vision add-on takes an image URL
The Cloudinary AI Vision add-on answers questions about an image through the Analyze API. The image does not need to be in the Cloudinary account.23

Who does it suit?

It suits a site or an app that already delivers its images and videos through Cloudinary. A check can start when an asset arrives in a folder, and a saved result can make an approved asset public. It is one of the few products here that checks the spoken words and the on-screen text of a video.

If your code must fetch or change the files, The Brand API Register describes the Cloudinary API.

When is another product a better fit?

  • If your brand check is about your own logo, your palette, and your typefaces, look at Bynder or Rainbrand. The logo rules here name well-known logos, the color rule checks the background, and the pages document no font check.
  • If an agent must start the check on work that it made, note that the documented API reads saved results. The pages document no call that starts a check.
  • If your work is documents or long text, look at Markup AI or Writer. Cloudinary Moderation works on images and videos.

Where are the vendor's own pages?

Sources

  1. cloudinary.com: cloudinary moderation checked November 16, 2026
  2. intercom.help: image rule types and how to use them checked November 16, 2026
  3. cloudinary.com: moderation checked November 16, 2026
  4. intercom.help: video rule types and how to use them checked November 16, 2026
  5. intercom.help: what is cloudinary moderation checked November 16, 2026
  6. intercom.help: set up ongoing moderation checked November 16, 2026
  7. intercom.help: run one time moderation checked November 16, 2026
  8. intercom.help: take action on moderation results checked November 16, 2026
  9. intercom.help: create a rule checked November 16, 2026
  10. intercom.help: understand moderation results checked November 16, 2026
  11. intercom.help: one time or ongoing moderation which should i use checked November 16, 2026
  12. intercom.help: image moderation checked November 16, 2026
  13. intercom.help: video moderation checked November 16, 2026
  14. intercom.help: what are rules checked November 16, 2026
  15. intercom.help: what are policies checked November 16, 2026
  16. intercom.help: i can t find or use a rule type checked November 16, 2026
  17. cloudinary.com: pricing checked November 16, 2026
  18. intercom.help: why wasn t my content caught checked November 16, 2026
  19. intercom.help: understand moderation actions checked November 16, 2026
  20. intercom.help: review a moderation run checked November 16, 2026
  21. cloudinary.com: cloudinary llm mcp checked November 16, 2026
  22. cloudinary.com: moderate assets checked November 16, 2026
  23. cloudinary.com: cloudinary ai vision addon checked November 16, 2026