Acceptable Use Policy
Last updated October 28, 2026
SnapKoi ("Platform," "we," "us") provides AI image editing and generation. This Acceptable Use Policy applies to every prompt, uploaded photo, and image the Platform makes or stores, and it is part of the Terms of Service. SnapKoi is an independent product. It is not affiliated with, endorsed by, or maintained by the providers of the models it uses, including OpenAI. Output is AI-generated.
Allowed. You can edit a product or scene photo you own or have permission to use (a new background, color, lighting, or style) and generate an original poster from a text prompt.
Not allowed. NSFW, explicit, sexually suggestive, and harmful content is not allowed. Illegal content is not allowed. Deepfakes and face swaps of real people are not allowed. This site has no face-swap tool.
1. Scope
This policy covers:
- Images: AI-generated or AI-edited images.
- User inputs: prompts and instructions you submit.
- Uploads: photos you send so an edit can be made.
The Platform does not generate audio or video, and it has no public gallery.
2. Prohibited content
The Platform prohibits generating the following six categories. Each one is prohibited on its own:
- Pornography / NSFW content.
- Violence / gore.
- Hate speech.
- Child-unsafe content (CSAM).
- Deepfakes / impersonation of real people.
- Content infringing copyright or trademarks.
2.1 Absolutely prohibited
The following is acted on as soon as we confirm it, and may be reported to authorities:
- Child sexual abuse material (CSAM) or any sexual content involving minors.
- Content that promotes or facilitates terrorism, extremism, or mass violence.
- Technical instructions for weapons of mass destruction.
- Content inciting genocide or ethnic hatred.
- Deepfakes of real individuals, including face swaps, nudification, intimate-scene compositing, and fabricated statements or actions, whether or not the individuals consent. The editor has no face-swap, undress, kiss, or hug tool, and asking for one is a violation.
2.2 Not offered
Adult content is not a restricted category we allow with age verification. It is prohibited. Violence and gore are prohibited, including when the prompt claims an artistic or educational purpose. We also do not provide medical, legal, or financial advice, and you must not present Platform output as that advice.
2.3 Other prohibited conduct
- Generating or spreading misinformation, including a real person saying or doing something they did not.
- Impersonating a real individual, brand, or organization.
- Infringing intellectual property, publicity, or privacy rights.
- Passing AI-generated output off as a photograph a person took, or as human-made, when you are asked.
- Bulk-generating spam or deceptive commercial content.
- Jailbreak prompts written to bypass these checks.
2.4 Photos of real people
You may upload only your own photo, or a photo for which you have explicit permission from the people in it. Uploading another person's photo or voice without their consent is prohibited. Before an edit is submitted, the editor asks you to confirm that the photo is yours or that you have permission. Permission does not authorize the deepfake content prohibited in section 2.1.
3. How content is reviewed
We run an automated check, a human review path for reports, and a periodic look at the blocklist.
3.1 Automated review
- Input: before a credit is reserved and before the image model is called, the prompt is sent to Waffo Prompt Sift (
POST /v1/actions/verification/scan-prompt). Only a verdict ofallowcontinues.blockandreviewdo not generate an image and do not charge a credit. - Output: after the model returns an image, and before that image is shown, the prompt is scanned again with the same API. The image is withheld unless that second verdict is
allow. Waffo does not publish an image-byte scanner; this text scan is the output check. The site has no public gallery, so published works are not scanned after the fact. - A local blocklist, reviewed 2026-10-28, also rejects NSFW, CSAM, violence, hate, self-harm, deepfake, and well-known copyrighted-character wording the classifier happened to allow. Prompt Sift is the list that stays current. The local list is a supplement and is updated when those terms change.
- Safety tool: Waffo Prompt Sift, with semantic scoring set to enforce. If Waffo Pancake is not the configured moderation provider, the same fail-closed rule applies to the provider that is configured.
3.2 Human review
Ray Cui, an individual based in China, reads reports sent to support@mossimageedit.com and through the report page. There is no separate 24-hour moderation vendor. An automated block is a machine decision; a report or an appeal is a human review of that decision. Moderation verdicts (the action, the reason code, and the request id, not the prompt text) are kept for 90 days.
3.3 Periodic review
At least once a month the operator reviews the local blocklist against the six categories above and a sample of stored moderation verdicts, and updates the blocklist when evasion terms show up. We do not commission a third-party audit.
3.4 Severity
| Level | Examples | What we do |
|---|---|---|
| L1 Critical | Child safety, terrorism | Automatic block, then human confirmation, then a ban and a report to authorities when the law requires it |
| L2 High | Deepfake of a real person, NSFW, copyrighted character | Automatic block. On a confirmed report, credits for that account can be zeroed and further use refused |
| L3 Medium | Other policy violations | Warning by email and removal from the response (the image is not shown and is not stored) |
| L4 Low | A first minor miss | Warning when someone reports it |
4. Reporting
Report content or behavior that breaks this policy, a child-safety concern, or a result that impersonates you or uses your work. Reporter identity is used only to reply and is not published. False reports are recorded and can themselves be acted on.
5. Where to report
- Report form: /report
- Email: support@mossimageedit.com
Those are the channels a person reads. There is no phone line and no in-app chat.
6. How fast we respond
| Level | First response | Decision |
|---|---|---|
| L1 Critical | As soon as the inbox is seen, and within 24 hours | Within 24 hours of confirmation |
| L2 High | Within 1 business day | Within 3 business days |
| L3 Medium | Within 3 business days | Within 7 business days |
| L4 Low | Within 5 business days | Within 15 business days |
The inbox is monitored by one operator in China (UTC+8), not a round-the-clock team. If a report needs longer, we email the reporter. The path is: received, graded L1–L4, automatic record plus human review, decision, enforcement, reply.
7. What we can actually do
- Warn you by email and ask you to stop.
- Refuse to generate, and not show an image. Finished images are not kept in our database, so there is no gallery to take down.
- Zero the credits on the email account and refuse further generations or purchases for that email.
- Permanently refuse access for L1 violations or repeated L2 violations.
- Report criminal L1 content to the authorities and cooperate with a lawful request.
8. Appeals
If a prompt was blocked or credits were removed and you think that was wrong, email support@mossimageedit.com within 14 days. Include the time of the request. The appeal is a human review of the automated decision and the moderation record. We aim to reply within 10 business days. A successful appeal restores unused credits that were removed in error.
9. Changes
Material changes are posted on this page. Where a change tightens a rule you already rely on, we aim to post it at least 15 days before it applies. Continuing to use the Platform after the update means you accept it.
10. Contact
- Safety, reports, and appeals: support@mossimageedit.com
- Operator: Ray Cui, an individual based in China
A postal address is given to Waffo Pancake during merchant review and is not published on this page. This policy is the public content policy for the Platform. It is not legal advice.