Choose what Comment Guard should allow, hide, delete or leave for review. Your keyword rules take priority; AI categories and sensitivity guide separate decisions, with actual outcomes recorded for inspection.
Required account permissions, recipient eligibility and Meta limits still apply. A matching event or public reply is not proof of a delivered DM.
Choose a keyword, its match behavior and an allow, hide or delete action. These rules take priority over the AI category decision.
The classifier evaluates categories such as spam/promotion, competitor links and abuse; confidence and your settings determine the next step. It can misclassify.
Hiding and deletion are different actions. A deleted comment is not guaranteed recoverable; a failed API request does not mean the action happened.
Inspect safe, pending, hidden or deleted outcomes and the recorded reason. Borderline link-only cases can remain up for review.
Configure a keyword action for a phrase you do not want, then test its match and inspect the result.
Do not assume every link is spam. Review the classifier’s outcome before taking irreversible action.
Check actual hide/delete outcomes and failures instead of assuming the comment section is fully filtered.
The custom keyword rule runs before the classifier. If no rule matches, the AI verdict is compared with the chosen sensitivity and category action. Low-confidence and borderline link cases may be left visible or pending.
The current sensitivity thresholds differ by setting; detection can miss unwanted content or misclassify a genuine question. An API refusal, missing permission or billing issue can stop the action. Review the event instead of assuming all spam disappeared.
The product supports configured hide and delete actions; “auto-hide only” was inaccurate. Do not promise that deletion can be reversed or that all moderation actions are recoverable.
Moderation does not establish a guaranteed ordering barrier for every reply or DM job. Verify moderation and automation outcomes independently; keep classification claims separate from what Instagram accepted.
No. Owner settings can select allow, hide or delete behavior. A delete action can be irreversible; review the actual outcome.
No. Keyword rules and AI categories differ; borderline link-only cases can be left visible or pending for review.
No. Confidence, sensitivity and category settings affect action, and false positives or missed comments are possible.
No guaranteed moderation-before-reply barrier is established. Check moderation and send outcomes separately.
Pricing · All features · How it works · Blog