Acceptable Use Policy

• Pornography and NSFW: sexual, pornographic or sexually exploitative content, including content involving animals.
• Violence, gore and cruelty: graphic injury, abuse, animal cruelty or celebration of violence. Gentle, non-graphic remembrance and decorative Día de Muertos skull motifs are permitted.
• Hate: attacks, dehumanization or incitement against people or groups based on protected characteristics.
• Child safety: child sexual abuse or exploitation material, sexualized minors, grooming or other content that endangers children, including synthetic material.
• Deepfakes and impersonation: synthetic depictions of real people or public figures, face or voice impersonation, fabricated speech or conduct, including where a person has consented. This is a pet-only product.
• Non-consensual intimate imagery (NCII): intimate imagery without consent, synthetic undressing, sexualized manipulation, threats to distribute such material, or tools to create it.
• Copyright and trademark infringement: unauthorized photographs, film scenes, characters, competitor media, music, logos or branding, or misleading affiliation claims.

Terrorism, violent extremism, mass violence, weapons of mass destruction, fraud, harassment, doxxing, unlawful activity, attempts to evade safeguards, and instructions that facilitate these harms are also prohibited.

Each uploaded photo must receive a manual decision tied to its exact file hash before a generation request can be submitted. Outputs are stored in a private quarantine. The reviewer must inspect the entire video and any audio before approving those exact bytes for account playback and download. Unreviewed, rejected or unverifiable outputs are withheld. Supplier acceptance uncertainty is also held for investigation; a request is not automatically submitted again.

The operator uses an authenticated local review tool through our existing Cloudflare authorization. Decisions, reasons, reviewer and content hashes are logged. The tool can reject an asset, revoke delivery access, and suspend or restore account access. We have not integrated Prompt Sift, an automated NSFW classifier, or an automated image/video moderation service. There is no round-the-clock moderation team.

Do not attach child sexual abuse material, intimate imagery, passwords, unnecessary identity documents or other sensitive material. A report link is provided with each video in the account library. Reports are reviewed by river through this support mailbox; this is not an emergency service. Make honest reports. We limit access to report information, but cannot promise absolute confidentiality where investigation or legal obligations require disclosure.

When river discovers a credible severe violation, access or delivery is blocked first while it is investigated. Actions may include refusal before generation, withholding or revoking a result, a warning, or account suspension for severe or repeated violations. Already downloaded copies cannot be recalled. A provider task already accepted may still finish; blocking our delivery does not claim to cancel upstream processing.

A confirmed generation failure returns the credits actually spent on that job once. Rejecting an undeliverable paid output also returns its spent credits. An unused review allocation is not a purchase; an allocation already sent to a supplier is not automatically restored. Cash refunds are handled separately under the Refund Policy. Limited review evidence may be retained for security, disputes or legal duties, and suspected unlawful material may be reported where required.