Trust and Safety Protocols
Last Updated: 9/10/2026
This page documents how we keep prohibited activity off Cosmix. Our Acceptable Use Policy says what is prohibited. This one says what we do about it, in the order it happens: what has to be true before an account can reach anybody, how we find violations, who decides, and what happens to the account.
1. Prevention, before an account can reach anyone
Cosmix is built so that most of what an abuse policy has to catch elsewhere cannot happen here at all.
Adults only. An account cannot be created with a date of birth under 18. The date of birth is not decoration on a dating profile here, it is the basis of the astrological chart the whole product runs on, so every account carries one and it is checked.
A verified email address. Required before an account is usable.
A verified face. Every dating account must pass identity verification: the person submits a live selfie, and it is matched against the photograph on their profile by AWS Rekognition at a high similarity threshold. Until that passes, the account cannot reach the match feed, cannot see saved matches, and cannot open a conversation. The only routes left open while unverified are the ones needed to fix or replace the profile photo.
One photograph, and it is a face. The only image anyone can put on Cosmix is the profile photograph that verification is matched against. There is no gallery, no album, and no second photo.
No images between users. Conversations carry text and the app's own structured cards. There is no attachment, no photo message, and no way to send an image to another person. The channel that carries explicit imagery on other platforms does not exist here.
Matching is not photo-first. People are drawn together by chart compatibility, and a card leads with an aura and a score rather than with a body. That is a product decision rather than a safety control, but it shapes who the platform attracts.
2. The rules people agree to
Every user accepts our Terms of Service at registration, which incorporates the Acceptable Use Policy. The policy names each prohibited category rather than gesturing at "objectionable content": trafficking and commercial sexual services, sexual exploitation and coercion, anything involving a minor, sexual violence, sexually explicit content, harassment and hate, and fraud.
3. Detection
Reporting, from wherever the problem is. Every profile card and every conversation carries a Report control, on both the mobile apps and the web app. The reasons offered include sexual or explicit content, child safety or exploitation, an underage user, and trafficking or solicitation, alongside harassment, fake profiles, and impersonation. A report and a block are one action: reporting somebody removes them from your feed and your conversations immediately, before any review has happened.
Blocking. Anyone can block anyone at any time, from the same two places. A block is immediate and mutual, and it notifies our moderation team.
Automated analysis of every report. A report is not a queue entry that waits for someone to open it. Each one is analyzed automatically as it arrives, against the conversation and the reported profile, and produces a category, a confidence score, and the specific evidence it relies on. That analysis is attached to the report before a person ever sees it.
Abuse signals. Requests are rate limited per client, and account, payment, and usage anomalies are visible to us independently of any report.
4. Review and decision
Every report reaches a person. Whether or not automated analysis produced a finding, every report raises an alert in our moderation channel, with the reason, the evidence, and what was decided.
The most serious categories are never closed automatically. Reports of sexual or explicit content, child safety or exploitation, an underage user, or trafficking or solicitation are held for human decision by design. Automation may gather the evidence and may act to contain the account immediately, but it cannot dismiss one of these. A person closes it.
Both sides are examined. The analysis reads who sent what. Where the person who filed the report is the one who broke the rules, that is surfaced and acted on, so reporting cannot be used as a weapon.
Within 24 hours. We act on reported content within 24 hours, as our Terms commit us to.
5. Enforcement
Enforcement is graduated, and it starts with containment because containment is reversible and delay is not.
- Removal from discovery. An account under a substantiated finding is taken out of matching immediately, so nobody new can reach it while the case is decided.
- Content removal. Offending content is removed.
- Permanent ban. A banned account cannot log in, is excluded from every match pool, and is excluded from re-engagement. Bans are permanent, and evading one is itself a violation.
- Preservation and referral. Where conduct may be criminal, the account and its contents are preserved and referred, as described below.
6. Child safety
We do not wait to be asked. Where we find child sexual abuse material or conduct that endangers a minor, we ban the account, preserve it and its contents, and report to the National Center for Missing and Exploited Children (NCMEC), which forwards to the appropriate authorities. Reports involving a minor are prioritized above all other work. Our Child Safety Standards Policy sets this out in full.
7. Working with law enforcement
Our Law Enforcement Guidelines document how officers reach us, what legal process each class of record requires, how we handle emergency disclosure where there is a risk of death or serious physical injury, and how we honor preservation requests.
8. Records
Every report is retained with its reason, the evidence gathered, the decision, and who or what made it. That record is what lets us answer a law enforcement request, review a decision, and see patterns across accounts rather than one report at a time.
9. Keeping these protocols current
Abuse changes, so these protocols do too. We revise them as new patterns appear rather than on a fixed schedule, and the date at the top of this page is when they last changed.
10. Contact
Safety matters, including anything on this page: safety@cosmixpairing.com. Law enforcement: see the Law Enforcement Guidelines. Anything else is on our contact page.