Acceptable use policy

Effective date: October 9, 2026

This policy applies to prompts, dialogue or voice direction, optional reference images, generated videos and any content shared through the service. It supplements the Terms of Service.

1. Six prohibited content categories

You may not use the service to generate, request, upload or distribute:

1. Pornographic or NSFW content: explicit sexual acts, nudity intended to be sexual, sexual exploitation or sexually explicit audio or video.
2. Violence or gore: graphic injury, gore, dismemberment, torture, glorification of real-world violence or instructions for carrying out violence.
3. Hate speech: content that attacks, dehumanizes or encourages violence against a protected group or person because of a protected characteristic.
4. Child sexual abuse or unsafe content involving minors (CSAM): any sexualized depiction, exploitation or grooming of a minor. This is zero tolerance.
5. Deepfakes or impersonation: deceptive content that impersonates a real person, or uses a real person's likeness without authorization, including non-consensual intimate imagery. The fixed Rumpelstiltskin lead is a fictional character; that does not authorize uploads of real people.
6. Copyright or trademark infringement: unauthorized copying of protected characters, logos, brands, footage, music, voices or a living creator's distinctive style.

2. Zero-tolerance content

We immediately block or remove CSAM, terrorism or extremist content, instructions for weapons of mass destruction, incitement to genocide or mass violence, and non-consensual sexual deepfakes. We may preserve relevant records and report suspected illegal activity to law enforcement where legally required or appropriate.

3. Additional prohibited behaviour

• fraud, scams, phishing, malware or instructions for wrongdoing;
• attempts to bypass safety checks, rate limits, payment controls or access restrictions;
• bulk scraping, spam, resale of access or training a competing AI model on our inputs or outputs; or
• presenting AI-generated content as human-created in a way intended to deceive, or using it as professional medical, legal or financial advice.

4. Model disclosure

The MVP uses MiniMax H3 Max Reference-to-Video through fal.ai. The request enables the provider safety checker. DeepSeek is used separately only for optional text prompt optimization. We are an independent product and are not endorsed by MiniMax or fal.ai.

5. Content moderation

Our moderation process uses several checks:

• Input screening: prompts submitted for optimization or video generation are checked against safety rules for the prohibited categories before they are sent to a provider. Requests that match a prohibited category are rejected before credits are reserved.
• Provider safety controls: video requests use fal.ai's safety checker. A provider rejection or failed safety check does not produce a preview; reserved credits are returned under the Refund Policy.
• Human review: reports and automatically flagged cases are reviewed by the operator. We do not claim that automated checks catch every violation, so users should report suspected violations.
• Periodic review: we review reported cases and safety-rule performance at least monthly and update the rules when new abuse patterns appear.

Generated videos are AI-generated and may contain mistakes. Review the result before sharing it and follow any disclosure obligations that apply where you publish it.

6. Reporting channel

Report a suspected violation, unauthorized likeness, copyright or trademark issue, child-safety concern, or harmful output by emailing 1848599547@qq.com with the subject Safety report. Include the public or private preview link if available, the reason for the report and any relevant order or job ID. Do not attach illegal sexual material; describe it and provide the location instead.

Reports are kept confidential where possible and reviewed impartially. Deliberately false or abusive reports may themselves lead to account action.

7. Response targets

• Child safety, terrorism, credible threats or non-consensual intimate imagery: initial review target within 2 hours and action target within 24 hours.
• Other high-risk reports: initial review target within 1 business day and action target within 3 business days.
• Copyright, trademark, privacy and general policy reports: initial review target within 3 business days and action target within 7 business days.

These are review targets, not a guarantee. We may request additional information or extend a review when needed to investigate safely.

8. Enforcement and appeals

Depending on severity, confidence and prior history, we may warn the user, reject a prompt, remove or withhold a result, restore or suspend credits, limit features, temporarily suspend an account, permanently terminate an account, or report suspected illegal activity. Severe or repeated violations may be acted on without a warning.

If you believe your content or account was incorrectly restricted, appeal by emailing 1848599547@qq.com within 14 days of the notice. We aim to complete an appeal review within 10 business days, where practical, by someone who did not make the original decision.

9. Contact

Safety reports, appeals and content-rights requests: 1848599547@qq.com.