Please note that this feature is still in Beta, it's not yet available for GA.
AI Image Moderation Beta helps communities identify potentially violating images and GIFs and route them into existing moderation workflows.
This feature is designed to support moderators, not replace them. AI Image Moderation can help flag content that may need review, but moderators continue to make the final decision.
Availability
AI Image Moderation is currently available as a beta feature.
To participate, customers must complete the beta process and sign the required beta agreement.
Interested in joining the beta? Use this form:
AI Image Moderation Interest Form
AI Image Moderation may not be available on all plans and may have an additional cost. Contact your CSM or Support for availability.
Data Processing
During the beta, images and GIFs may be processed in the United States or Europe.
Canada-based clients should be aware that there is no Canada-based processing path. Their image and GIF content will be processed in Europe.
How AI Image Moderation Works
When enabled, AI Image Moderation scans images and GIFs submitted by users. If content is flagged, it is routed into the community’s moderation workflow based on the selected enforcement mode.
AI Image Moderation may identify content categories such as nudity, violence or disturbing content, hateful or offensive content, substances, and other configured violation types.
Moderators can review flagged content and take the appropriate action.
Enable AI Image Moderation
Admins can turn AI Image Moderation on or off from the feature settings.
When AI Image Moderation is disabled, the related configuration options are hidden.
Enforcement Mode
Enforcement Mode controls what happens while image and GIF content is being checked.
Remove If Violating
In Remove If Violating mode, posts publish first. If AI Image Moderation flags an image or GIF, the post may be removed and routed into moderation.
This means the content may be visible briefly before removal. The exact timing can vary.
Moderators can review the moderation record and restore or approve the content if it was incorrectly flagged.
Use this mode when your community is comfortable allowing posts to publish first, with flagged content removed if needed.
Hold Until Cleared
In Hold Until Cleared mode, posts with images or GIFs are held before publishing while the scan runs.
Users see this message: "Your post will appear after it is approved."
If the image or GIF is cleared, the post can publish. If it is flagged, the post remains unpublished until a moderator reviews it.
Use this mode when your community would rather pause image and GIF posts before they appear publicly.
Sensitivity Level
Sensitivity Level controls how strict AI Image Moderation is when checking images and GIFs.
Available levels include Low, Medium, High, and Very High.
Lower sensitivity may allow more content through. Higher sensitivity is stricter and may create more false positives.
Violation Types
Admins can choose which violation types AI Image Moderation should check for.
Use the options in this setting to align AI Image Moderation with your community’s moderation needs.
Moderator Review
AI Image Moderation uses existing moderation workflows.
Depending on the enforcement mode and outcome, flagged content may appear in Reports or Escalations.
Moderators can review the AI signal, including the violation type and severity, then take normal moderation actions.
Moderator actions may include approve, restore, reject, remove, edit before approval, assign, escalate, and other available moderation actions.
DSA options require Community Management Dashboard. DSA-specific moderation options are available only when the newer Community Management Dashboard (CMD) and DSA mode are enabled. These options are not available through the legacy moderation features. If your community is still using the legacy experience, contact Vanilla Support or your CSM about availability and migration.
DSA Behavior
The workflows below apply only to communities using the newer Community Management Dashboard with DSA mode enabled. Communities using legacy moderation features do not have access to these DSA options.
AI Image Moderation works with existing DSA moderation flows.
The key distinction is whether the content was ever published.
DSA disabled + Remove If Violating: The post publishes first, then may be removed if flagged. The moderation item appears as an Escalation.
DSA disabled + Hold Until Cleared: The post is held before publishing. The moderation item appears in Reports as Approval Required.
DSA enabled + Hold Until Cleared: The post is held before publishing. If rejected, no DSA removal notice or appeal applies because the content was never published.
DSA enabled + Remove If Violating: If published content is removed, the regular DSA removal notice, notification, and appeal flow apply.
For more information, see Digital Services Act Compliance.
Frequently Asked Questions
Is AI Image Moderation generally available?
No. AI Image Moderation is currently a beta feature.
How do I join the beta?
Use this form: AI Image Moderation Interest Form
Customers must complete the beta process and sign the required beta agreement.
Is AI Image Moderation free?
AI Image Moderation may not be available on all plans and may have an additional cost. Contact your CSM or Support for availability.
Where are images and GIFs processed?
Images and GIFs may be processed in the United States or Europe. Canada-based clients should be aware that there is no Canada-based processing path. Their image and GIF content will be processed in Europe.
Does AI Image Moderation remove content automatically?
AI Image Moderation can flag content and route it through moderation workflows. Depending on the selected enforcement mode, content may be removed after publishing or held before publishing. Moderators can review flagged content and override false positives when needed.
What is the difference between Remove If Violating and Hold Until Cleared?
Remove If Violating allows posts to publish first. If an image or GIF is flagged, the post may be removed and routed into moderation.
Hold Until Cleared holds posts with images or GIFs before publishing until the scan clears them or a moderator reviews them.
Can moderators override false positives?
Yes. Moderators can review flagged content and approve, restore, reject, edit, assign, escalate, or take other available moderation actions.
What should I choose for sensitivity?
Lower sensitivity may allow more content through. Higher sensitivity is stricter and may create more false positives. Choose the level that best matches your community’s moderation standards.