Categories
Findings are sorted into hate, NSFW, violence, profanity, political, legal, public feud, brand criticism, substance and deceptive content. Each finding shows the post, the caption or transcript excerpt, the frame where relevant, and when it was published.
Severity
Critical and high findings are the ones to read first; the headline count is of critical and high findings after our review. Medium and low findings are context — swearing, mild content — and are listed below.
Reviewed vs unreviewed
Every detection goes through a second review pass that confirms or refutes it. A finding that could not be reviewed is labelled unreviewed and never outranks a reviewed one. If you believe a finding is wrong, use Report as false positive on it — that feeds the detector.
Legal and accusatory matters carry their outcome
A finding about a lawsuit, charge or allegation always states how it ended — charges declined, case dismissed, settled, convicted, or explicitly unresolved as of a date. A finding we could not carry to its outcome is not published. We would rather omit a finding than publish half of one.
Deleted content
Where web archives hold posts the creator has since removed, they appear in their own section, dated, with the archive source.
Very large public figures
For household-name public figures the report reads the public record around the person — press, legal and regulatory matters, public discourse, commercial history — plus a text pass over their written posts, instead of a frame-by-frame scan of a million-post catalogue. The report says which mode ran.
Still need help? Message us and a person will answer.