AI Image Moderation helps protect researchers and stakeholders from seeing offensive or inappropriate images when viewing reports that contain image uploads from participants.
Researchers and other stakeholders must be confident that they can review and share image-based research without offensive content appearing in their reports. AI Image Moderation is designed to increase researchers willingness to share reports internally, since images are screened, and researchers don't need to manually moderate every image.
When enabled, Alida automatically screens images using an AI service before they appear in Modern Reporting. This page explains what that means, what is and is not moderated, and how to turn the feature on or off.
As with any AI tool, Alida AI can make mistakes.
What is AI Image Moderation?
AI Image Moderation is an optional feature that uses Amazon Web Services (AWS) Rekognition to automatically analyze images for potentially inappropriate content.
For more information on the image moderation service we use, see What is Amazon Rekognition?
What types of content are moderated?
At a high level, the service is designed to detect categories such as:
- Nudity and sexual content
- Sexual situations or partial/explicit nudity
- Graphic or non‑graphic violence
- Gore, physical injury, or self‑harm
- Potentially disturbing or shocking imagery
When AI Image Moderation is turned on, Alida outsources the technical moderation of images to a third-party provider. Therefore, Alida cannot:
- Make exceptions for specific images.
- Explain individual moderation decisions, for either blocked or permitted images.
Note: Alida Technical Support does not have access to additional details about why a specific image was blocked for violating a moderation rule by the AWS Rekognition service.
- Customize or override these categories.
How do I turn image moderation on or off for my organization?
Your organization can choose whether or not to take advantage of AI Image Moderation. To turn AI Image Moderation on or off, please contact your Customer Success Manager.
The current status of AI Image Moderation is clearly displayed in Image Upload tiles.
| On | Off |
|---|---|
|
|
|
Current limitations of AI Image Moderation
Moderation is only applied to images displayed on screen in the report.
The maximum sizes for images uploaded by survey participants are:
- 60 MB for modern surveys
- 20 MB for classic surveys
Both of these limits exceed the maximum size for images to be processed by the AWS Rekognition service used by AI Image Moderation to process the images. The limit depends on the geographical region where your organization's data resides:
- 5 MB for na1.alida.com
If you are on na1, your URL for accessing the Alida platform is https://app.na1.alida.com/
- 15MB for all other regions