Effective Date: 2026-07-29 Last Updated: 2026-09-17
This policy describes how we moderate photos, profiles, messages, and behaviour on courtme - what we check, what we remove, what happens to an account, and how to challenge a decision. What is and is not allowed is set out in our Community Guidelines. How to report someone, and what blocking does, is in our Reporting & Blocking Policy.
| Tier | Trigger | What happens |
|---|---|---|
| Allow | Passes every automated check | Published immediately |
| Flag | Possible violation, low confidence | Sent to a human within 24 hours |
| Block | High-confidence violation, e.g. explicit imagery or weapons | Blocked at once, you are told, appeal available |
| Emergency | Imminent threat to someone's safety, child sexual abuse material, credible threats of violence | Account suspended within 1 hour, escalated internally and to law enforcement where appropriate |
Our tools are AWS Rekognition for image content and AI classifiers for text, with human moderators for context, ambiguity, and appeals. Rekognition also runs a face-detection quality check on the photo - is there a usable face, is it sharp and well lit. The face measurements themselves are discarded immediately: since 11 September 2026 no face geometry is stored against your account. What we keep about a photo is the moderation outcome: whether it passed, which of our rules it hit, how many faces were seen, and a capped summary of what the classifiers reported, including any text detected in the image. We do not run face recognition, and we do not match a selfie to an ID.
Automated systems make mistakes in both directions. That is why every automated removal can be appealed to a person (Section 6).
We review both content that is reported to us and content our systems flag on their own. Automated classifiers run across photos and text; a report is one way into the surrounding context, including the relevant thread.
Your messages are stored on our servers and are not end-to-end encrypted, which is what makes this review possible. We bound it with controls rather than promises:
See Section 3.6 of our Privacy Policy for the full description of internal access.
| Situation | Action |
|---|---|
| First minor violation | Warning and the content comes down |
| Repeated violations | Temporary suspension, 24 hours to 7 days |
| Serious violation | Permanent ban |
| Zero-tolerance violation (Community Guidelines §7) | Immediate permanent ban, plus a law enforcement referral where appropriate |
We can ban without a prior warning where the conduct is severe or the pattern is clear. A banned account cannot be replaced by a new one; evading a ban is itself a violation.
If you are in the EU, you may also have out-of-court dispute options under the Digital Services Act; contact legal@courtme.ai and we will point you to them.
courtme IRL events are run by independent hosts, not by us. A host has limited authority over their own event only:
Removal from one event is not a finding against you. If you think a host removed you unfairly, email appeals@courtme.ai.
We require valid legal process - a subpoena, warrant, or court order - for non-public account data, except where there is an emergency involving a risk to someone's life or safety. We file NCMEC reports for child sexual abuse material as US law requires. We may publish aggregate transparency numbers (reports received, share actioned, categories) with no personal data in them.
We update this policy as the product and the law change. Material changes are posted here with a new effective date.