Part 1: The SI8 Rights Verified Framework
Standing rules for all represented content and production work.
What SI8 Will Work With
- Human-directed content: Meaningful human creative authorship documented (not just prompt entry)
- Commercial licenses: All AI tools used with paid commercial plans, receipts on file
- Approved tools only: Tools on Approved or Caution tier list (no Prohibited tools as primary generation source)
- Documented provenance: Complete Chain of Title for every piece of content
The No List — What SI8 Won't Work With
Hard boundaries that protect filmmakers, clients, and SI8 from highest-risk areas:
- ❌ Celebrity Likeness: Real people's faces without consent
- ❌ Voice Cloning: Real person voices without authorization
- ❌ IP Imitation: Copyrighted characters, brands, trademarks
- ❌ Deepfakes: Deceptive content pretending to be real
- ❌ Political Content: Persuasion or advocacy content
- ❌ CMI Stripping: Watermark/attribution removal (DMCA violation)
- ❌ Prohibited Tools: Veo consumer tier, tools in active litigation
- ❌ No Human Authorship: Prompts alone (not copyrightable per USCO)
Every SI8 project comes with a documented rights position. Brands get legal defensibility, not just creative output.
Part 2: Legal Landscape (Current State)
Copyright, liability, regulatory compliance — US, Taiwan, Singapore, EU
Copyright Ownership
USCO January 2025 Human Authorship Requirement: The U.S. Copyright Office requires documented human authorship for AI-generated works. Prompts alone do not establish copyright ownership. SI8's Chain of Title Field #9 (Version History) captures the editorial paper trail required.
U.S. Copyright Office, "Copyright and Artificial Intelligence, Part 2: Notice of Inquiry and Request for Comments" (Jan 2025). Human creative expression must be meaningfully present.
Damage Mitigation
73% Statutory Reduction via Innocent Infringer Defense: Chain of Title documentation supports innocent infringer defense under 17 U.S.C. § 504(c)(2), reducing minimum statutory damages from $750 to $200 per work and preventing willful penalties up to $150,000.
Regulatory Compliance
Colorado AI Act "Rebuttable Presumption" Safe Harbor: Colorado SB24-205 explicitly rewards documented systems with legal presumption of reasonable care. Chain of Title = statutory compliance infrastructure, avoiding $20K penalties per violation.
EU AI Act Article 50: Phased transparency obligations for synthetic/deepfake content (2026+). SI8's Model Disclosure (Field #2) satisfies EU transparency requirements.
FTC Operation AI Comply: Substantiation requirements for AI-related marketing claims. Chain of Title provides audit trail for FTC compliance.
Training Data Liability
User liability is real: Tool providers' ToS generally disclaim liability for copyright infringement. Buyers assume risk. Tool tier classifications matter — using tools in active litigation (Midjourney, Stability AI) increases exposure.
SI8 does not work with tools under active studio litigation as primary generation source. Prohibited tier exists to protect filmmakers and buyers.
DMCA CMI Preservation
No watermark stripping: DMCA § 1202 prohibits removal of copyright management information (CMI). Watermark removal is a separate violation from copyright infringement. SI8 requires CMI preservation.
Part 3: Tool Tier Classifications
Not all AI tools carry equal risk. SI8 classifies tools into risk tiers based on training data transparency, litigation status, and commercial terms.
| Tool | Tier | Rationale |
|---|---|---|
| Adobe Firefly | ✅ Approved | Adobe Stock-trained models only. Indemnification for Firefly outputs (exclusions apply). |
| Runway (paid) | ✅ Approved | Best commercial terms. Service improvement clause documented. Paid plan required. |
| Pika (paid) | ✅ Approved | Clear commercial use authorization. Paid plan required. |
| Kling (paid) | ✅ Approved | Approved commercial plan. Note: retains sublicensable license to use outputs for AI training — flag in Chain of Title where exclusivity matters. |
| Kling | ⚠️ Caution | Retains permanent sublicensable license to use outputs for AI training. Approved for catalog but flagged for exclusivity deals. |
| Veo (consumer) | ❌ Prohibited | Consumer tier = no commercial use. Enterprise tier may be evaluated in future. |
| Hailuo / Minimax | ❌ Prohibited | Active Disney/WB lawsuit (Nov 2025). Too high risk for commercial use. |
| Midjourney, Stability AI | ❌ Prohibited | Active studio litigation. SI8 will not use as primary generation source until litigation resolves. |
Adobe indemnifies Firefly only. Excludes Runway, Kling, Pika, Midjourney, any modifications or combinations. Creative teams want best-in-class tools (Runway, Kling), not just safe tools (Firefly). SI8's positioning: "Adobe = safe but boring. SI8 = best creative tools made safe through Rights Verified vetting."
Part 4: Platform Policies
YouTube
- Disclosure requirement: "Altered or synthetic content" must be disclosed if realistic people, places, or events depicted
- Label placement: Video description or on-screen notice
- Enforcement: Demonetization, recommendation suppression, or removal for non-disclosure
TikTok
- Disclosure requirement: AI-generated content toggle required at upload
- Label: "AI-generated" label appears on video
- Enforcement: Content removal for non-disclosure
Meta (Facebook, Instagram)
- Disclosure requirement: "Made with AI" label for photorealistic content
- Detection: Meta auto-detects some AI content; manual disclosure also accepted
- Enforcement: Reduced distribution for non-compliance
iQIYI (China/Taiwan Streaming)
- Content requirements: Full rights documentation required for licensing
- AI disclosure: Currently evaluating policies (as of Feb 2026)
- SI8 advantage: Chain of Title satisfies documentation requirements
- No formal AI disclosure requirement yet (as of Feb 2026)
- Best practice: Proactive disclosure in post text
SI8's Chain of Title = platform-compliant delivery. Model Disclosure (Field #2) + Tool Provenance (Field #1) satisfy disclosure policies across all major platforms.
Part 5: Vetting Criteria (7 Categories)
Pass/fail thresholds for each review category. For full vetting criteria with examples, see Rights Verified Overview.
1. Tool & Plan Verification
Pass: All tools on Approved or Caution tier list. Paid plan receipts on file for every tool. Commercial license status confirmed.
Fail: Prohibited/High-Risk tool as primary source; missing receipts for any tool.
2. Human Authorship Evidence Quality
Pass: Declaration describes specific iteration decisions, editorial choices, post-generation editing. Minimum 150 words, substantive.
Fail: Generic ("I wrote prompts and selected outputs") with no specifics. AI made all meaningful creative decisions.
3. Likeness & Identity
Pass: All faces clearly synthetic and non-identifiable. All voices original or AI-generated with no specific real-person identity.
Fail: Any identifiable real-person face or voice without consent. Any doubt = fail.
No exceptions. No gray area. If a face is possibly identifiable, it fails. This is the highest legal exposure category.
4. IP & Brand Imitation
Pass: No character/creature/logo/brand element recognized as belonging to real IP holder. No replication/parody of real brand trade dress.
Fail: Clear imitation of copyrighted character. Protected trade dress replicated. Trademarked logos/identifiers visible.
5. Brand Safety
Pass: Suitable for mainstream commercial brand attachment, no restrictions. OR has specific documented restrictions in Chain of Title.
Fail: Graphic violence, sexual content, political persuasion, drug use (celebratory/instructional), any No List content.
6. Audio & Music Rights
Pass: All audio is original AI-generated using paid commercial plans (receipts on file). All licensed audio has documentation. Work is silent (no audio).
Fail: Unlicensed commercial music without documentation. Audio license doesn't cover commercial use or specific territory/medium.
7. Modification Rights (Tier 2 Gate)
Authorized — Full Work: All scenes eligible for product placement regeneration.
Authorized — Specific Scenes: Only listed scenes eligible for product placement.
Not Authorized: Tier 1 (as-is licensing) only. Filmmaker can update authorization anytime.
Part 6: Legal Risk Reduction Mechanisms
How SI8's Chain of Title reduces AI content liability through four quantifiable legal mechanisms.
Mechanism 1: Damage Mitigation
73%+ Statutory Damage Reduction
Chain of Title documentation supports innocent infringer defense under 17 U.S.C. § 504(c)(2), reducing minimum statutory damages from $750 to $200 per work and preventing willful penalties up to $150,000.
Supported by: Fields #1 (Tool Provenance), #3 (Rights Verified Sign-off), #4 (Commercial Use Authorization), #9 (Version History)
Mechanism 2: Regulatory Compliance
Colorado "Rebuttable Presumption" Safe Harbor
Colorado AI Act explicitly rewards documented systems with legal presumption of reasonable care. Chain of Title = statutory compliance infrastructure, avoiding $20K penalties per violation.
Supported by: Fields #2 (Model Disclosure), #3 (Rights Verified Sign-off), #6 (Category Conflict Log), #9 (Version History)
Mechanism 3: Copyrightability Protection
USCO Human Authorship Requirement
U.S. Copyright Office (Jan 2025): Prompts alone don't grant ownership. Chain of Title Field #9 documents the editorial paper trail required for your brand to actually own the copyright.
Supported by: Field #9 (Version History — captures human authorship evidence)
Mechanism 4: Distribution Enablement
E&O Insurance + Platform Compliance
Chain of Title = chain-of-title equivalent for AI content. Satisfies E&O underwriting requirements and platform disclosure policies (YouTube, TikTok, streaming).
Supported by: All 9 fields (comprehensive documentation)
- Damage Reduction: 73% minimum ($750 → $200 statutory damages)
- Regulatory Penalties Avoided: Colorado $20K, NY $1K-$5K per violation
- Distribution Enablement: E&O insurance documentation, platform compliance
- Copyrightability Protection: USCO human authorship requirement met
Ready to Get Your Content Verified?
$499 per video. 90-minute expert review. 5-business-day turnaround.
Submit for Verification →