Get answers to common questions about copyright.sh, AI content licensing, and creator rights protection.
copyright.sh is a content licensing platform that enables fair compensation for creators when their content is used to train AI models. Think of it as "ASCAP/BMI for the AI age" - we ensure creators get paid when AI companies use their work.
We provide a simple meta tag system for content protection, real-time usage tracking with HMAC verification, and automatic payments to creators when AI companies license their content.
Traditional licensing requires complex negotiations, legal paperwork, and often takes months. copyright.sh provides:
• Instant setup with a simple meta tag
• Transparent pricing set by creators
• Real-time payments (daily vs quarterly)
• Global coverage (not limited to specific regions)
• 90% revenue share to creators vs ~60-70% in traditional licensing
Yes, copyright.sh is built on established copyright law. Content creators own the copyright to their work and have the legal right to license it on their terms.
We're also supporting 35+ ongoing lawsuits against AI companies that use content without permission. Our system provides the legal documentation and proof needed for copyright enforcement.
For Creators: Free to join. We take a 10% platform fee—you keep 90% of every license.
For AI Companies: Pricing follows a 2-axis model based on jurisdiction, risk tier, and content class. Rates start at $0.05 / 1K tokens for open-web public-domain–equivalent text and scale up to $50 / 1K tokens for embargoed newsroom or premium book content (reflecting recent News Corp precedents).
No minimum commitments, monthly fees, or hidden charges—you pay only for the exact tokens, bars, seconds, or frames you train on.
Getting started takes just 5 minutes:
1. Sign up for a free account
2. Add our meta tag to your website: <meta name="ai-license" content="rate=$0.05; terms=https://copyright.sh/terms; id=YOUR-ID">
3. Set your rates and start earning when AI companies use your content
Earnings depend on your content quality, traffic, and rates. Based on our launch projections, creators can expect:
• Tech bloggers: $85-$140/month
• Researchers: $200-$312/month
• News sites: $150-$250/month
• Authors: $50-$400/month
Use our earnings calculator on the pricing page for a personalized estimate.
Popular rate ranges by content type:
• News articles: $0.001-$0.005 per 1K tokens
• Blog posts: $0.01-$0.03 per 1K tokens
• Technical docs: $0.04-$0.06 per 1K tokens
• Research papers: $0.05-$0.10 per 1K tokens
• Books/premium content: $0.05-$0.10 per 1K tokens
You can adjust rates anytime based on demand and results.
You get paid daily via Stripe once you earn $10 or more. Payments are automatic - no invoicing or paperwork required.
We support multiple currencies (EUR, USD, GBP) and provide detailed earning reports for tax purposes.
Yes, you can opt out anytime by removing the meta tag from your website. There are no contracts or commitments.
You'll receive any pending payments, and AI companies will no longer be able to license your content through our platform.
Commercial AI training is likely not fair use. With 35+ active lawsuits seeking billions in damages, the legal risk is enormous.
Licensing provides:
• Legal protection from copyright claims
• Higher quality curated training data
• Brand protection - avoid "AI theft" accusations
• Regulatory compliance for EU AI Act and similar laws
Integration takes under an hour:
1. Register for API access
2. Check licenses before using content via our API
3. Use licensed content in your training with automatic billing
We provide SDKs for Python, JavaScript, and Go, plus comprehensive API documentation.
Our pricing is still trivial compared to litigation exposure:
• Open-web text: $0.05 – $0.50 / 1K tokens
• Professional & journalistic: $1 – $5 / 1K tokens
• Embargoed newsroom & premium books: $10 – $50 / 1K tokens
Training 10 M tokens on premium news therefore costs ~$500K—still a rounding error next to nine-figure statutory damages (e.g. The New York Times' $1 B+ claim).
Yes, we provide comprehensive compliance documentation:
• HMAC-verified usage logs with cryptographic proof
• Licensing agreements for each piece of content
• Payment receipts showing creator compensation
• Compliance certificates for regulatory audits
• Real-time dashboards for your legal team
This is actively being litigated, but the trend is clear: courts are skeptical of AI companies' fair use claims for commercial training.
Key factors working against fair use:
• Commercial nature: AI companies make billions from training
• Market harm: AI output can replace original creators
• Substantial copying: Entire articles/books used for training
• No transformative purpose: Goal is to reproduce similar content
robots.txt is not legally binding and many AI companies ignore it. It's a courtesy convention for web crawlers, not a copyright protection mechanism.
Our meta tag system provides:
• Legal enforceability based on copyright law
• Licensing terms clearly stated
• Payment mechanisms built-in
• Usage tracking for enforcement
We enforce licensing through multiple mechanisms:
• API blocking: Non-paying AI companies can't access content
• Legal documentation: Clear evidence for copyright claims
• Collective action: Supporting 35+ ongoing lawsuits
• Public transparency: Naming companies that refuse to pay
• DMCA takedowns: For unauthorized usage
Copyright protection is recognized internationally through treaties like the Berne Convention. Our system works globally because:
• Universal copyright: No registration required in most countries
• International enforcement: DMCA and equivalent laws worldwide
• EU AI Act compliance: Strict requirements for training data provenance
• Cross-border payments: We handle multi-currency transactions
We are fully aligned with the EU AI Act (Art. 53 provenance), the U.S. NO FAKES Act draft, and Australia's proposed mandatory licensing scheme. Our policy team tracks 20+ global bills and updates the license generator automatically.
Yes. We have three provisional patents filed covering (1) bar-second metering for generative audio, (2) cryptographic audit trails for AI training data, and (3) dynamic jurisdictional pricing. Final filings are scheduled for Q3 2025.
HMAC (Hash-based Message Authentication Code) provides cryptographic proof that content was licensed:
1. AI company requests license for specific content
2. We generate HMAC signature using SHA-256 with content URL + timestamp + token count
3. Usage is logged with tamper-proof signature
4. Creators get paid based on verified usage
This prevents fraud and ensures accurate billing for all parties.
A token is roughly 3-4 characters or about 0.75 words. For example, "Hello world!" is about 3 tokens.
We count tokens using the same methods as major AI companies:
• Text content: Standard tokenization (similar to GPT models)
• Average webpage: ~450 tokens
• Blog post: 800-2,000 tokens
• Research paper: 5,000-15,000 tokens
We use multiple detection methods:
• API integration: Ethical AI companies check licenses before using content
• Web crawling patterns: We monitor for training-specific access patterns
• Model output analysis: Looking for training data "leakage" in AI responses
• Community reporting: Creators can report suspected unauthorized use
Our system works with protected content too:
• Meta tag in public areas: Add to login pages, headers, or preview content
• API integration: For sites with APIs that AI companies might access
• Retroactive protection: If your content appears in training data, we can help with licensing claims
The key is making your licensing terms discoverable by AI companies.
Phase 1 focuses on web-first, meta-tag licensing with verified API metering. Full search/RAG and in-context licensing will arrive in Phase 2 (Q4 2025) once the creator corpus exceeds 1 Bn licensed tokens.
Our team is here to help. Get personalized answers to your questions about copyright.sh.
Contact Support