Acceptable Use Policy & AI Content Safety
Last Updated: November 20, 2026
This Acceptable Use Policy ("AUP") defines the mandatory content safety standards, acceptable usage boundaries, and enforcement rules for ZeroClip (zeroclip.app).
By accessing ZeroClip or submitting any media or text to our clipping pipeline, you agree to comply with this policy and our Terms of Service.
1. Prohibited Content & Activities
ZeroClip enforces a strict zero-tolerance policy against unlawful, harmful, and abusive content. You may not upload, process, transcribe, reframe, or generate clips containing:
- Child Sexual Abuse Material (CSAM) or any form of child exploitation or endangerment.
- Non-Consensual Intimate Imagery (NCII), sexually explicit media, non-consensual sexual violence, or intimate-scene compositing.
- Deceptive Impersonation & Deepfakes: Fabricating malicious representations of real persons without consent, political deepfakes, or non-consensual face modifications.
- Hate Speech & Harassment: Content promoting violence, hatred, discrimination, or severe harassment against individuals or protected groups.
- Self-Harm & Dangerous Content: Material promoting suicide, self-harm, cyberattacks, weapons manufacturing, or illegal drug operations.
- Intellectual Property Infringement: Uploading materials without the necessary rights, licenses, or authorizations from copyright owners.
2. Multi-Stage Content Moderation & Safety Architecture
To ensure platform safety and compliance with international Merchant of Record (MoR) standards, ZeroClip operates automated content moderation at multiple processing stages:
- Pre-Generation Prompt & Input Screening (Waffo Prompt Sift): All user-submitted text inputs, topics, and titles are screened in real time via Waffo Prompt Sift (
POST /v1/actions/verification/scan-prompt) before AI models are invoked. Generation is permitted only when the check yields anallowverdict. Prompts flagged for restricted or harmful content are rejected immediately. - Media Ingestion & Face Alignment: Visual speaker tracking relies on automated computer vision without generative face synthesis, ensuring source video integrity without altering personal facial identities.
- 72-Hour Automatic Data Purge: All uploaded source footage and exported render outputs expire and are permanently deleted from Cloudflare R2 storage within 72 hours.
3. Underlying AI Model & Technology Disclosure
ZeroClip transparently discloses the underlying artificial intelligence and computational models powering our media processing engine:
| Pipeline Stage | Technology / Model Provider | Purpose |
|---|---|---|
| Speech-to-Text & Word Timestamps | OpenAI Whisper / Faster-Whisper | Audio transcription and word-level subtitle timecode synchronization |
| Speaker Detection & Reframe | Google MediaPipe | Spatial bounding-box speaker face detection and vertical 9:16 layout tracking |
| Highlight & Viral Hook Scoring | MiniMax M3.1 Flash / Advanced LLM | Semantic transcript analysis and viral highlight timecode selection |
| Media Composition | FFmpeg | Video clipping, subtitle burning, and 1080p MP4 encoding |
4. Enforcement, Account Suspension & Credit Forfeiture
When an account or submission is identified as violating this AUP:
- The infringing job is aborted immediately, and temporary media artifacts are purged.
- ZeroClip reserves the right to suspend or permanently ban the user account and associated IP ranges.
- As outlined in Section 4 of our Terms of Service, any remaining minute credits or balances are permanently forfeited without refund.
- Egregious violations involving child safety or imminent harm will be reported to competent law enforcement authorities.
5. Reporting Violations & Inquiries
If you detect any content processed through ZeroClip that violates this policy, please notify our safety team:
- Designated Safety Email: contact@zeroclip.app
- Response Commitment: All safety and abuse notices are reviewed by human staff within 48 business hours.
- Related Documents: Terms of Service · Privacy Policy · DMCA Policy · Contact Us