After a compliance scan, every finding needs a decision: fix it, dismiss it, or tell Warrant it isn't relevant. This article explains what a finding shows, why some findings are dismissed automatically, how to add a flag the scan missed, and how your team's dismissals and relevancy feedback are used. It's for anyone who reviews scan results, and for admins who set up review settings.
New to scanning? Start with Scan an Asset for Compliance.
What a finding shows
Open a file, click Scan, and look at the findings list. Findings are grouped by regulator or policy source (for example a federal regulator, Disclosures, or Brand Guidelines). Each finding card shows:
Risk level: a High, Medium or Low badge with a matching icon.
Where it is: a page number for documents and PDFs, or a timestamp for audio and video. Click the card to jump to that spot in the file.
AI badge: every scan finding is generated by Warrant's AI. Hover over the badge to read: "This finding was generated by Warrant's AI system. A human reviewer makes the final decision."
Source: the policy or rule that was triggered, a description of the issue, and, when available, a View source regulation link, a Required Disclosure box with the copy that should appear, or a View Disclosure link. The Required Disclosure box always shows approved wording from your disclosure. When the file's channel, audience or product matches one of the disclosure's variants, it shows that variant.
Findings a reviewer added by hand carry a Reviewer badge instead. See Add a flag the scan missed below.
Warrant also judges how confident it is that each finding applies to your content. Depending on your settings, findings it is less sure about can be dismissed automatically, as explained below.
How confidence and Review sensitivity work together
Your admin's Review sensitivity setting decides how confident Warrant must be before a finding is shown. See Global Sensitivity Control for Compliance Scans for how to change it. The levels run from Flexible, which shows the fewest findings, through Mild (the default), Balanced and High, to Maximum, which shows the most. Higher sensitivity shows more lower-confidence findings.
A few things to know:
Brand guideline findings can use their own level. If your admin sets Brand guidelines sensitivity to a level other than Same as compliance sensitivity (default), brand guideline findings are compared against that level instead. All other findings use Review sensitivity.
Your counts only include open findings. The High, Medium and Low counts on the Scan tab leave out dismissed findings, whether the system or a person dismissed them.
Relevancy Confidence Auto-Dismissal
This switch decides what happens to findings that fall below your sensitivity level. It's on by default.
On: when a scan runs, findings below your sensitivity level are dismissed automatically and marked System Dismissed. They don't count toward the compliance score and don't appear in the active list, but you can still review them (see below).
Off: no finding is dismissed because of low confidence. Every finding stays open for a person to review and counts toward the score.
The switch applies to scans run after you change it. Findings that were already System Dismissed stay dismissed until someone restores them. It doesn't affect dismissals based on reviewer feedback, described in How relevancy feedback is used below.
To change it:
Click Settings in the left sidebar.
Under Compliance, select Review settings.
In the AI compliance section, turn Relevancy Confidence Auto-Dismissal on or off. It appears when Enable AI compliance scanning is on, and changes save straight away.
Only admins can change this setting.
System Dismissed vs. Dismissed
A dismissed finding is grayed out and carries one of two labels:
System Dismissed: Warrant dismissed it automatically when the scan ran. The card reads Auto-dismissed by system with the date, and explains why, for example that its confidence was below your sensitivity level, or that reviewers previously marked its policy not relevant. Nobody has reviewed these findings.
Dismissed: a person dismissed it. The card shows who dismissed it, the date, and their reason if they gave one.
View dismissed findings
Above the findings list, click Dismissed. The number in brackets shows how many findings were dismissed in this scan. You can also click the gray Dismissed count next to the risk counts under the score.
Click Active to see only open findings, or All to go back. All shows open findings and findings a person dismissed. System Dismissed findings only appear under Dismissed.
Dismiss a finding
If a finding doesn't need a fix, for example because the issue is handled elsewhere, you can dismiss it.
On the finding card, click Dismiss.
Optionally, enter a Reason for dismissal. A reason isn't required, but it helps approvers and auditors understand your decision.
Click Dismiss to confirm.
The finding is removed from the compliance score, and the file's score updates straight away. Dismissing doesn't tell Warrant the policy was wrong. To do that, use relevancy feedback, described below.
Restore a dismissed finding
Dismissed findings, whether dismissed by a person or by the system, have two options:
Undismiss reopens the finding. It counts toward the compliance score again. No feedback is recorded.
Mark Relevant also reopens the finding, and records a "relevant" vote from you on it. Warrant asks you to confirm first. For a System Dismissed finding, the message reminds you that it was never reviewed and that marking it relevant records your review as an affirmative human judgement. Use this when the system was wrong to dismiss something.
A restored finding stays open in the findings list, in the risk counts, in Copy results and in Export to spreadsheet, even if its confidence is below your sensitivity level. It stays that way unless someone dismisses it again.
Add a flag the scan missed
If you spot an issue the scan didn't catch, you can add it to the scan yourself. It is then reviewed and recorded like any other finding.
Open the file and click Scan.
Above the findings list, click Add flag.
In the Add a flag window, enter a Title.
Under Policy, choose the policy the issue falls under, or No policy covers this.
Choose a Severity: High, Medium or Low.
Optionally enter the Page (for documents and PDFs) and Details about what's wrong and where.
Click Add flag.
What happens next:
The flag appears in the findings list with a Reviewer badge, so it is easy to tell apart from AI findings. It counts toward the compliance score like any other open finding, and you can dismiss it or comment on it in the same way.
It is recorded in the file's history on the Audit tab, with who added it and when.
A policy task is raised, and Super Admins are notified, so an admin can decide whether a policy needs to change. Adding a flag never changes a policy by itself.
If the same version is scanned again, your flag is kept on the new scan. When a new version is added, flags from the earlier version are listed in the Versions panel under Reviewer flags to re-check, so you can confirm whether the issue was fixed.
Add flag is offered on the latest completed scan of the version you're viewing. It isn't available while a scan is running.
Comment on a finding
Click the speech bubble icon on a finding to open its comment thread, type in Add a comment… (@ to mention), and send. The icon shows how many comments a finding has. Use comments to discuss a finding with your team before deciding what to do with it.
Mention a teammate: type @ and pick someone who can open the file. They get a notification, in-app and by email depending on their Mentions setting.
Who else is notified: people who commented on or were mentioned in the finding's thread before, and the person who uploaded the file, depending on their Comments setting. Nobody is notified about their own comment.
Resolve and reopen: hover over a comment and click the check mark (Resolve) once it has been addressed. The comment shows a Resolved tag. Click Reopen (the circular arrow) if it needs more attention. Anyone who can comment on the file can resolve and reopen comments, and each change is recorded with who made it and when.
Edit or delete: hover over your own comment and click the pencil (Edit) or trash (Delete) icon. Workspace admins can edit or delete anyone's finding comment. Edited comments are marked (edited). Warrant asks you to confirm a delete, and the comment is then shown in the thread as Comment deleted.
Nothing is lost: when a comment is edited, its earlier wording is kept, and when it is deleted, its text and history are kept for audit. Other readers see only the current text or Comment deleted. Workspace admins can review the full history with Export comment audit on the file's Audit tab.
On the Notifications page, a finding comment notification has a View finding button that opens the file on the Scan tab at that finding's thread. See Adding Comments in the File Viewer.
Give relevancy feedback
Relevancy feedback tells Warrant whether a finding was actually relevant to your content. It's separate from dismissing: you can give feedback on a finding whether or not you dismiss it.
On the finding card, click the thumbs up (relevant) or thumbs down (not relevant) icon.
In the Mark as Relevant or Mark as Not Relevant window, optionally add a note explaining why. When you mark a finding not relevant, you can also pick a reason: Doesn't apply to this product, channel or location, Rule doesn't apply to our business, Disclosure is already present, Misread the content or Duplicate of another finding. A reason is optional. Click a selected reason again to clear it.
Click Relevant or Not Relevant.
A Relevant or Not Relevant tag appears next to the icons. Each person has one vote per finding. Voting again replaces your earlier vote.
Picking a reason helps Warrant tell a policy that doesn't apply to your content apart from one that applied but wasn't broken. The same reasons are offered when you mark a finding not relevant in Studio and Monitor.
If your admin turns on Show Relevancy Feedback on Flags (in Settings under Compliance, Review settings; off by default), the notes reviewers write appear on the finding itself, the way a dismissal reason does. When it's off, the notes are visible to admins only, on the Policy audit trail page and in the How it's doing section of each policy's page, which also counts the reasons reviewers picked.
How relevancy feedback is used
Votes are counted per policy across your whole company, not per file. They're only used for your company's scans, and they affect future scans, not findings that already exist.
Policy tasks: when a policy is repeatedly marked not relevant, Warrant creates a task suggesting an admin review that policy, and Super Admins are notified. The votes, reasons and notes are on the policy's page under How it's doing. See Policy Tasks and the Policy Audit Trail.
Higher scrutiny: repeated not relevant votes on a policy make Warrant apply it more strictly, so it is flagged only when there is clear evidence of a violation. The Scan feedback summary shows a banner listing these policies.
Automatic dismissal: if more than one reviewer marks a policy not relevant, on several different files, and not relevant votes keep outweighing relevant ones, new findings from that policy can be dismissed automatically and labelled System Dismissed, with a note that reviewers previously marked it not relevant. This happens even when Relevancy Confidence Auto-Dismissal is off. High-risk findings are never dismissed this way, so they always stay open for a person to decide.
Relevant votes, including those from Mark Relevant, count against not relevant votes. If a policy keeps getting dismissed but your team finds it useful, marking its findings relevant brings it back. If a policy really doesn't fit your business, an admin can edit, narrow where it applies, lower the risk level of, or disable it on the AI Policies page.
Where it's recorded
On the finding: who dismissed it, when, and the reason, or why the system dismissed it.
Policy audit trail: every thumbs up, thumbs down and Mark Relevant vote appears in the Policy audit trail with the person, the policy and any note. Filter by Feedback to see only these entries, and click Download CSV to export them.
The file's audit trail: thumbs up and thumbs down votes, flags added by reviewers, and comments added, edited, deleted, resolved or reopened on findings, also appear in the file's history on the Audit tab.
Comment audit export: workspace admins can click Export comment audit on the file's Audit tab to download a CSV of every finding comment and Notes comment on the file, including deleted comments and earlier wording, with who made each change and when.
Policy tasks: each flag a reviewer adds raises a task, and Super Admins are notified. See Policy Tasks and the Policy Audit Trail.
Who can do what
Dismiss, Undismiss, Mark Relevant and Add flag need the Review permission.
Thumbs up and thumbs down need the Relevancy Feedback permission.
Commenting on findings needs the Comment permission and at least Can comment access to the file. With view-only access you can read a finding's comments but not add to them. Resolving and reopening comments needs the same access.
Authors can edit and delete their own finding comments. Workspace admins can edit and delete any finding comment, and are the only people who can export the comment audit.
Changing Review sensitivity, Relevancy Confidence Auto-Dismissal or Show Relevancy Feedback on Flags needs admin access.
If you don't see these buttons on a finding, ask your admin to check your role. See Managing Roles & Permissions.