Review integrity findings
Read, triage, and act on the collusion evidence the platform gathers during your event.
While your event runs, the platform reviews its own activity once an hour and raises a finding whenever it sees a pattern worth a human look.
Findings are evidence, not verdicts. Nothing here changes a score, bans a player, or notifies anyone. Every consequence is yours to apply.
Integrity is available on Starter and above; see Plans and limits.
Read the list
Open Trust & Safety → Integrity findings in your event dashboard. The list opens filtered to Open findings; the other review states are a click away. Each row shows, among other columns:
- Subject: the team or player the finding is about, with a one-line plain-language summary.
- Check: the identifier of the check that raised it. Quote it when you cite a case or ask us about one.
- Severity: how much attention it deserves, as Critical, High, Medium, or Low.
- Confidence: how much the evidence carries on its own.
- Confirmed: the pattern has no innocent explanation.
- Probable: a strong pattern with a narrow innocent explanation.
- Suspect: a real signal that ordinary play also produces, so treat it as corroboration, never a conclusion on its own.
- Events: how many separate occurrences sit behind the finding.
Filter by review state, by severity, or by Re-escalated: findings you already dismissed that have since picked up new activity, so a case that grows after you closed it comes back rather than staying buried.
Above the table, the page shows when the checks last ran, and names any that are paused for want of data. Read that line before concluding an event is clean: an empty list and a check that never ran look identical otherwise. A paused check can also under-report, so treat its findings as provisional until the missing data arrives.
Open a finding
Clicking a row opens the detail sheet: the full summary, every recorded occurrence with its timestamps and the other party where there is one, and any other findings raised against the same subject. A team with one suspect finding is usually noise; the same team appearing across three unrelated checks is a pattern.
Move the finding through review as you work:
- Reviewing: you have picked it up.
- Actioned: you applied a consequence.
- Dismissed: you decided it is not a problem.
You can reopen anything. Your decision is never overwritten by a later check.
Act on a confirmed case
Consequences are applied from the ordinary participant pages, not from Integrity:
- Adjust points: open the team or player from Players and apply a signed penalty with a reason. This is the proportionate response to most confirmed cases.
- Ban the player: from the player detail sheet, with an optional expiry. The reason is shown to them at sign-in.
- Remove or move a player: solves and points stay with the original team.
Deleting a player removes their account and every submission, solve, and hint unlock attached to it. The finding itself remains, but the evidence behind it is gone, and so is the evidence behind anyone else’s finding that names them; neither finding can be reviewed or defended afterwards. Apply the consequence you want first, and delete last if at all.
Exclude your own accounts
Staff accounts that test challenges produce the same patterns as cheating accounts, so they generate findings you don’t want to wade through.
Add them under Trust & Safety → Internal & test teams with a team handle or a username and a reason. No findings are raised for anything on that list from the next hourly check. They still appear in other teams’ evidence, so excluding your own staff team does not blank it out of a genuine case elsewhere. Remove an entry and findings resume.
Judgement, not proof
Findings are evidence, not verdicts. Corroborate before you act: a subject appearing across several unrelated checks is a far stronger case than a single finding. Where a case is close, prefer the proportionate response. A points adjustment you can reverse is recoverable; a deleted account and its evidence are not.