Who can use this feature?
|
This feature is activated on-demand only for now. Contact your Client Success Partner (CSP) if you’re interested.
When a comment is published and is considered harmful by AI, the following happens:
- After 10 seconds, the comment is unpublished.
- An email is sent to the creator of the comment, explaining that the comment is under review.
- An email is sent to the admins of the group where the comment was published, as well as to platform admins, allowing them to unarchive the comment.
Definition of harmful comments
We define "harmful comments" as "text content including language that promotes violence, hate, illegal activities, misinformation, self-harm, exploitation, or harassment, causing physical, emotional, or societal harm".
Direct, blunt, or negative criticism of a product or service is considered safe, as long as it does not include profanity or insults.
About AI accuracy
AI moderation is designed to minimize errors, but false positives can occur. This means that occasionally, a comment that is not harmful may be archived. The estimated false positive rate is 0.2%. Group and platform administrators can review and unarchive any comment flagged by mistake.