Volver al inicio

Imagine Dance Acceptable Use Policy — AI-Generated Content

Content standards, prohibited categories, moderation, reporting, and enforcement for the Imagine Dance Seedance AI video generation platform.

Última actualización: 2026-09-26

Imagine Dance Acceptable Use Policy — AI-Generated Content

AI-Powered Video / Image / Audio Generation Services

Last Updated: September 26, 2026

Imagine Dance ("Imagine Dance," "the Platform," "we," "us," or "our") is committed to providing safe, lawful, and responsible AI-generated content services. This Acceptable Use Policy ("Policy") applies to all content generated, published, distributed, or stored through the Platform, and sets out our content standards, moderation, reporting procedures, and enforcement actions. All users, developers, and merchants using the Platform must comply with this Policy.

This Policy complements Section 10 (Acceptable Use Policy) of the Imagine Dance Terms of Service — the Terms of Service points users at this full content policy; this is that policy.

1. Scope

This Policy applies to all content types, including:

  • Text: AI-generated text, dialogue, code, articles, captions, metadata, prompts, parameter settings, negative prompts, and workflow data.
  • Images: AI-generated or edited images, illustrations, reference-image uploads, first-frame / last-frame uploads, thumbnail previews, frame extracts, watermarked previews, and any cropped / upscaled / post-processed derivative.
  • Audio / Video: AI-synthesised voice, music, soundtracks, lip-synced footage, generated short-form videos, reference-video uploads, reference-audio uploads, and any edited, cropped, concatenated, re-rendered, or watermarked derivative of a generated video.
  • User Inputs (Prompts): prompts and instructions submitted by users — including but not limited to text prompts, negative prompts, reference-media metadata, parameter choices, workflow graphs, and form submissions. Prompts are content and are subject to this Policy in the same manner as Outputs.
  • Conduct through the Service: sharing links, publishing to external platforms, distributing generated results, curating public galleries of outputs, embedding outputs in third-party apps, or automating access through wrappers / bots / unofficial clients.

This Policy applies equally and without exception to every use of the Service through the web UI, the mobile web, or any automation / wrapper / bot that accesses the Service's endpoints.

2. Content Standards & Prohibited Categories

The Platform prohibits generating the following six categories of content:

  • Pornography / NSFW content.
  • Violence / gore.
  • Hate speech.
  • Child-unsafe content (CSAM).
  • Deepfakes / impersonation of real people.
  • Content infringing copyright or trademarks.

The clauses below break these six categories and other related violations down further.

2.1 Absolutely Prohibited — Zero Tolerance

The following will be actioned immediately and WILL be reported to authorities where legally required:

  • Child sexual abuse material (CSAM) or any sexual content involving minors — including content that sexualises, exploits, endangers, or harms minors (any person under the age of 18, or the local age of majority if higher).
  • Content that promotes or facilitates terrorism, extremism, or mass violence.
  • Technical instructions for weapons of mass destruction, explosive devices, or manufacture of firearms / ammunition.
  • Content inciting genocide or ethnic hatred.
  • Non-consensual deepfake sexual content targeting real identifiable individuals.
  • Violent animal-cruelty content, instructions for torture or mutilation of living beings, or self-harm / suicide tutorials.
  • Any violation of the UN Optional Protocol on the Sale of Children, Child Prostitution and Child Pornography (OPSC), US 18 U.S.C. §§ 2251–2252A, EU Directive 2011/92/EU, UK Protection of Children Act 1978, Singapore Penal Code, or equivalent applicable local law.

2.2 Restricted — Subject to Review

For the avoidance of doubt — because Imagine Dance is an AI video-generation platform intended for general-audience creative use, the following categories are prohibited, not "allowed with conditions." We list them here only because they appear as "restricted" in the upstream AUP template; no age-verification or special-compliance carve-out is offered for any of them on this Platform:

  • Adult content, pornography, erotic fetish, hentai, ecchi, or any sexualised / erotic content.
  • Graphic violent or gore content without a clearly non-exploitative educational, historical, or documentary creative context.
  • Politically sensitive, misleading, or election-interfering synthetic content during the 90-day window before and 7-day window after any national election, referendum, or plebiscite.
  • Unqualified medical / legal / financial advice presented as fact without the required professional disclaimers and licensing.

2.3 Generally Prohibited Conduct

  • Generating or spreading misinformation, proven-falsehoods, disproven claims, or deceptive synthetic content intended to mislead a reasonable viewer into believing it depicts real events, without a prominent, unremovable, legible, close-proximity textual disclosure reading "THIS IS AI-GENERATED SYNTHETIC CONTENT — IT DOES NOT DEPICT REAL EVENTS."
  • Impersonating real individuals, brands, organisations, public officials, law-enforcement officers, judges, election officials, or protected-category persons, without that individual or entity's explicit, uncoerced, written, informed, signed consent (a dated "model release" or equivalent document naming the specific Output or category of Outputs it covers).
  • Infringing intellectual property, publicity, privacy, moral rights, database rights, trade-secret rights, or trademark / trade-dress rights of any third party.
  • Deceiving users by passing AI-generated content off as human-created, or as the official work of a third-party brand, studio, publisher, artist, or broadcaster, in a manner likely to cause consumer confusion, passing off, or reputational harm.
  • Bulk-generating spam, deceptive commercial content, SEO keyword-farmed content, or counterfeit-goods promotional material.
  • Jailbreak prompts or prompt-injection attacks designed to bypass Platform safety measures, guardrails, keyword blocklists, classifiers, OCR filters, or the upstream Kie AI safety-alignment layer — including prompt obfuscation, token splitting, homoglyphs, Unicode injection, polyglots, or equivalent safety-evasion techniques.
  • Using any Input or Output to train, fine-tune, LoRA-train, distil, benchmark, evaluate, or dataset-curate a competing AI model — commercial or non-commercial.

3. Content Moderation

We operate a multi-layer moderation system: automated review, human review, and periodic audits. A moderation step runs on inputs AND outputs of every generation request — no generation is exempt.

3.1 Automated Review (real-time)

All requests and generated content are scanned in real time by safety filters.

  • Input filtering: keyword blocklists, semantic safety detection, negative-prompt validation, reference-image OCR, reference-video frame sampling, and prompt-intent classification on every submitted Input before a job is dispatched to the upstream model provider.
  • Output filtering: real-time post-generation safety assessment of every generated Output, including video frame sampling, thumbnail OCR, perceptual-hash matching against known CSAM / violence hash-sets, and classifier re-scoring.
  • Usage pattern analysis: monitoring for anomalous generation-pattern bursts, rapid bursts of high-risk-category prompts, repeated evasion attempts, and IP / account-level correlation across signups.
  • Safety models used: (a) the upstream Kie AI (Seedance 2.0 / Seedance 2.0-fast / Seedance 2.5) provider safety-alignment layer — which rejects jobs at inference time; (b) Imagine Dance's proprietary Prompt Sift keyword + classifier guardrails running pre-generation; (c) post-generation Output scanners and perceptual-hash matching.

3.2 Human Review

  • Auto-flagged content enters a human review queue and is triaged within the timelines in Section 6 (Response & Processing Timelines).
  • User-reported content is reviewed manually by the Trust & Safety team.
  • Review team: on-call moderation team during business hours, escalated to senior on-call for L1 Critical reports outside business hours.
  • Reviewers complete annual content-safety training and sign non-disclosure and conflict-of-interest agreements.

3.3 Periodic Audits

  • Random content sampling performed weekly.
  • Monthly evaluation and iterative improvement of classifier accuracy, false-positive / false-negative rates, and keyword-coverage gaps.
  • Third-party legal review of this Policy and enforcement procedures at least annually, or whenever material new AI-safety regulation enters force.

3.4 Content Severity Levels

| Level | Description | Review | Action | |---|---|---|---| | L1 — Critical | Child safety, terrorism, WMD, non-consensual sexual deepfakes, CSAM, animal-cruelty instructions | Auto-block + immediate on-call human confirmation | Immediate permanent ban + report to NCMEC / IWF / competent law enforcement | | L2 — High | Severe violations: violence-glorification, hate campaigns, intentional deepfake impersonation, second confirmed copyright strike | Auto-flag + priority human review within 24 hours | Content removed + 14-day account suspension | | L3 — Medium | General violations: single hate-speech occurrence, first non-egregious impersonation, first valid uncontested DMCA / takedown notice, adult-content generation attempt | Scheduled human review | Written warning + content removal | | L4 — Low | Minor violations: accidental policy breach, single uncopyrightable fair-use-style dispute, low-severity false-positive dispute, uncivil user conduct towards support | Report-triggered human review | Written warning + edit / take-down if required |

4. Reporting

4.1 When to Report

  • Content or behaviour violating this Policy's content standards.
  • Critical violations involving child safety, terrorism, WMD, or imminent harm to a real person.
  • AI impersonation or infringement of your own copyright, trademark, right-of-publicity, or privacy rights.
  • Suspected security vulnerabilities, account takeovers, unauthorised account access, or payment / billing fraud.
  • Any other situation the Platform's Trust & Safety team ought to be aware of.

4.2 Reporting Principles

  • Confidentiality: Reporter identity is strictly protected and is never disclosed to the accused user or any third party except as required by law or competent court order.
  • Impartiality: All reports receive independent, objective review; reviewers with a real or perceived conflict of interest are recused.
  • Anti-abuse: Malicious, vexatious, knowingly false, or bulk bad-faith reports are recorded and may result in reporting-user account restriction or termination.
  • Good-faith presumption: A single report is never taken as proof of a violation; a full investigation is conducted before enforcement action is taken, except for L1 Critical auto-block cases where delay would create an unreasonable risk.

5. Reporting Channels

| Channel | Address / Location | For these report types | |---|---|---| | Dedicated AUP-violation and trust-and-safety email | moderation@imaginedance.io | General AUP violations, content you believe violates Section 2 (all six categories), deepfake impersonation, L2–L4 severity | | DMCA / copyright / trademark / IP takedown email | dmca@imaginedance.io | Copyright infringement (DMCA notices and counter-notices), trademark / trade-dress complaints, right-of-publicity / personality-right infringement, Repeat Infringer Policy strikes | | Child-safety / CSAM / terrorism / WMD / life-safety emergency email | security@imaginedance.io (immediately forwarded to senior on-call + NCMEC CyberTipline / local IWF / competent law enforcement where required) | L1 Critical severity only — CSAM, non-consensual sexual deepfakes, terrorism, WMD, credible threats of imminent real-world harm | | Billing / refunds / subscription / credit / payment-fraud email | billing@imaginedance.io and refunds@imaginedance.io | Unauthorised charges, refund requests under Terms of Service Section 7, credit-expiration disputes, subscription lifecycle, suspected payment fraud on your account | | General platform support email (also displayed in the site footer) | support@imaginedance.io — THIS IS THE PRIMARY CUSTOMER-SERVICE CONTACT DISPLAYED IN THE SITE FOOTER. | All other enquiries, general questions, help with the web UI, feature requests, escalation of any other unresolved report, and cross-channel re-routing of misdirected reports to the correct specialist address |

For CSAM or life-safety emergencies, email security@imaginedance.io AND, if you are willing to identify yourself, include: (a) the exact URL or task ID of the offending Output; (b) why you believe it falls into L1 Critical; (c) your relationship to the matter (victim, parent / guardian, witness, researcher, platform trust-and-safety partner). Do NOT attach or re-distribute CSAM imagery; only include identifying metadata. We respond to L1 Critical reports within 2 hours and report confirmed CSAM to NCMEC (US CyberTipline) + the relevant local Internet Watch Foundation / national hotline + competent law enforcement, as required by law and our contractual reporting obligations with our upstream sub-processors Waffo Pancake, Kie AI, and Cloudflare.

6. Response & Processing Timelines

| Violation Type | Initial Response | Resolution | |---|---|---| | L1 — Critical | Within 2 hours | Within 24 hours, including any mandatory regulatory / law-enforcement reporting | | L2 — High | Within 24 hours | Within 3 business days | | L3 — Medium | Within 3 business days | Within 7 business days | | L4 — Low | Within 5 business days | Within 15 business days |

If extended processing is required (e.g., awaiting counter-notice windows, coordinating with upstream provider Kie AI on shared investigations, or obtaining independent legal advice), we will notify the reporter proactively with reasons.

Processing workflow:

  1. Received — auto-acknowledge (except anonymous L1 reports) with a ticket number.
  2. Triage — classify L1–L4, route to correct specialist reviewer, escalate to on-call if L1.
  3. Review — automated signals (hash matches, classifier scores, audit trail) + human review of Inputs, Outputs, surrounding context, and user history.
  4. Decision — either No Action, Warning, Content Removal, Suspension, Permanent Ban, or Mandatory Reporting. Written rationale for any Yes-action decision is preserved in the case record.
  5. Enforcement — action applied, content hash blocklisted, credit balances forfeited where applicable per Terms of Service § 13.2.3.
  6. Notification — accused user notified of the enforcement category, sections of this Policy they violated, duration if temporary, forfeitures if any, and appeal path.

7. Enforcement Actions

| Action | Measure | Applies to | |---|---|---| | 1. Warning | Written notice requiring removal, edit, or corrective action, logged on the user's account record | First-time L4 minor violation; first L3 medium non-infringement violation | | 2. Content removal | Violating content taken down; its perceptual hash blocklisted so it cannot be regenerated verbatim | Confirmed L3–L4; all L1–L2 | | 3. Feature restriction | Reduced daily quota, restricted model access, temporarily disabled public-link sharing, or generation rate-limit tightened | Repeat L4 minor; single L3 medium; safety-evasion attempt | | 4. Account suspension | Temporary account suspension pending completion of human review or the suspension period; no generations, no credit consumption, no refunds during the suspension window | L2 High; cumulative L3 x 2 | | 5. Permanent ban | Permanent termination of access to the Service; all sessions invalidated; all outstanding credit balances (subscription bundled, paid packs, promotional, manual grants — all) immediately and permanently forfeited as a breach remedy under Terms of Service § 13.2.3. No refund, no cure, no reinstatement except by a successful Section 8 appeal. | L1 Critical; repeat L2 High x 2; CSAM; non-consensual sexual deepfake; Repeat Infringer Policy third strike; malicious false reporting | | 6. Legal action | Preserve all logs, assets, Inputs, Outputs, session metadata, timestamps, IP data, and user-identifying information for the maximum period permitted by law; report to NCMEC / IWF / national CSAM hotline; cooperate fully with competent law-enforcement and regulatory investigation; notify upstream contractual sub-processors Waffo Pancake, Kie AI, and Cloudflare as required by our agreements with them. | Criminal-level L1: CSAM, terrorism, WMD, credible threats of imminent real-world harm |

The enforcement actions above are not exclusive — we may combine actions where appropriate (e.g., content removal + permanent ban + legal reporting for L1 CSAM). All enforcement actions, timelines, and forfeiture decisions are preserved in a case-log record for the maximum retention period required by applicable law.

8. Appeals

You may appeal an enforcement decision within 14 calendar days of receiving the enforcement-notice email. Appeals submitted after 14 days are not reviewed — no exceptions.

  • Appeal channel: Email moderation@imaginedance.io with the subject line APPEAL — [your account email] — [enforcement ticket number]. Attach: (a) the original enforcement notice you received (screenshot or full quoted text); (b) a clear, concise, evidence-based written explanation of why you believe the decision was incorrect; (c) any supporting evidence (written consent for deepfake / impersonation, proof of copyright ownership / license, fair-use legal analysis, signed model release, etc.).
  • Timeline: Appeals are reviewed within 10 business days of receipt (L1 Critical appeals are reviewed within 48 hours; note however that L1 Critical appeals involving confirmed CSAM cannot result in reinstatement).
  • Review method: Reviewed by a Trust & Safety reviewer independent of the original decision (the original reviewer is recused). The independent reviewer reviews the full case record, the automated signals, the human-review notes, and your appeal evidence, and renders one of three final decisions: (a) Upheld — original enforcement action stays in place; (b) Modified — enforcement action reduced in severity (e.g., permanent ban → 30-day suspension + credit forfeiture → partial credit reinstatement); or (c) Reversed — enforcement action cancelled, content restored (if technically feasible), any forfeited credits re-granted, account warning log entry cleared.
  • Finality: The appeal decision is final and internally non-appealable. Nothing in this Section limits your right to pursue any statutory or common-law remedy you may otherwise have against us.

9. Transparency & Policy Updates

  • We will publish a content-safety transparency report every calendar quarter, summarising (in aggregate, without identifying individual users): counts of L1–L4 reports received, most common violation categories, automated-vs-human actioned split, median triage and resolution times by severity tier, counts of appeals and outcomes, and counts of mandatory law-enforcement / NCMEC reports.
  • Material changes to this Policy are notified at least 15 calendar days in advance via the last-known email on file for registered users and by updating the "Last Updated" date at the top of this page.
  • Minor, non-material clarifications (grammar fixes, typo corrections, adding new example scenarios to existing prohibited categories, updating email addresses or third-party-provider names, aligning with new upstream Kie AI safety-alignment model versions without changing the substantive prohibited-category list) take effect immediately upon posting with no advance notice.
  • Your continued use of the Service after the effective date of an updated version of this Policy constitutes acceptance of the updates. If you do not agree to an updated Policy, your sole remedy is to cease using the Service and, if you have a positive credit balance, request a prorated refund of unused paid-credit-pack balances pursuant to Terms of Service Section 7.

10. Contact Us

| Role | Email Address | |---|---| | Safety / Security / L1 Critical reports | security@imaginedance.io | | AUP / Moderation / Trust & Safety (including appeals) | moderation@imaginedance.io | | DMCA / Copyright / Trademark / IP Takedowns | dmca@imaginedance.io | | Billing / Subscriptions / Credits / Refunds | billing@imaginedance.io and refunds@imaginedance.io | | General Platform Support (primary, footer-visible) | support@imaginedance.io |

Legal entity identity and registered address information is set out in the Imagine Dance Terms of Service — Section 15 (Governing Law, Dispute Resolution, and Contact Information).

This Policy is for general reference only and does not constitute legal advice. We recommend having it reviewed by a qualified AI-compliance attorney before you rely on it for your own compliance programme.