📌 1. Understanding AI-Generated Content in IP Law

AI-generated content includes:

Text (e.g., articles, stories)

Images, videos, and music

Software code

Designs and inventions

Key challenges:

Inventorship / Authorship: AI cannot be a legal author or inventor. Rights must accrue to a human or legal entity.

Ownership & Licensing: Multinationals need to structure licenses carefully to comply with jurisdictional IP laws.

Cross-border enforcement: Some jurisdictions have recognized AI-created works differently, affecting licensing terms.

📌 2. Copyright Considerations

🧠 Core Principle

Most countries require human authorship for copyright protection.

AI-generated works without human intervention may not receive copyright protection, limiting enforceability.

📍 Key Cases

(a) Naruto v. Slater (US, 2018)

Facts: A monkey took a selfie; the photographer claimed copyright.

Decision: The court ruled non-human authors cannot hold copyright.

Implication: Courts may treat AI-generated content similarly — only humans can be authors.

Licensing Takeaway: Only works with human creative input can be licensed as copyrighted material.

(b) Thaler v. Comptroller-General of Patents (UK, 2021)

While a patent case, the UKIPO highlighted: AI cannot own rights.

Implication: In copyright, human involvement is critical for licensing.

(c) US Copyright Office Policy on AI (2022)

USCO clarified: AI-generated works without human authorship are not copyrightable.

Licensing Tip: Document human creative contribution to ensure rights can be licensed.

📌 3. Patent Considerations

While AI-generated content often involves creative works, inventions and software derived from AI may qualify for patents if human inventors are named.

📍 Key Cases

(d) Thaler v. Vidal (US, 2022)

AI named as inventor; patent rejected.

Licensing Strategy: Only human-named patents can be licensed internationally.

(e) EPO DABUS Decisions (Europe, 2020–21)

AI inventor applications rejected.

Licensing Tip: Assign patents to a company or human; then license to other jurisdictions.

📌 4. Trade Secret Strategies

AI-generated algorithms, models, or proprietary datasets can be protected as trade secrets.

Advantages: No human authorship needed.

Challenges: Must maintain secrecy; enforceable through contracts.

Case Example:

(f) Waymo v. Uber (2017–2018, US)

Theft of AI-related trade secrets in self-driving technology.

Settlement reached: $245M, plus restrictions on Uber’s use of certain technology.

Licensing Takeaway: Trade secrets can be licensed across borders if confidentiality agreements are robust.

📌 5. Multinational Licensing Strategies for AI-Generated Content

Here’s a step-by-step strategic framework:

Step 1: Determine IP Eligibility

Copyright: Only works with human creative input.

Patents: Only inventions with human inventors.

Trade secrets: AI algorithms, training data, and models.

Step 2: Structure Ownership

Assign rights to a legal entity (e.g., company).

Human contributors sign assignment agreements to the company.

The company becomes licensor for all jurisdictions.

Step 3: Draft Licensing Agreements

Key clauses:

Rights Granted: Exclusive, non-exclusive, or limited field.

Territory: Specify country/region.

Royalties/Revenue Sharing: Especially if AI generates ongoing content.

AI Output Attribution: Clarify human involvement for enforceability.

Compliance & Updates: Adapt to local IP laws.

Step 4: Address Enforcement Risks

AI content can infringe third-party rights (training data, algorithms).

Include indemnity clauses and IP warranties.

📌 6. Notable Multinational Licensing Cases

(g) Getty Images v. Stability AI (2023, US)

AI trained on copyrighted images; Stability AI licensed outputs.

Ongoing litigation around copyright infringement.

Takeaway: Licensing AI-generated content may require securing upstream rights to training data.

(h) Microsoft Licensing ChatGPT (Global)

Microsoft licensed OpenAI technology for enterprise use globally.

Strategy: Central entity holds IP; sublicenses issued per jurisdiction.

Shows importance of centralized rights management for multinational licensing.

(i) Disney AI-Generated Art Licensing (US & EU)

AI-generated images of popular characters.

Licensing required human oversight to ensure copyright validity.

Key Point: Human review ensures the work is licensable internationally.

(j) IBM Watson Licensing (Global)

AI content generation via Watson platform.

IBM’s licensing included data rights, IP ownership, and AI output rights, with regional adjustments.

Lesson: Multinational licensing requires careful mapping of ownership, rights, and enforcement.

📌 7. Practical Takeaways

Always assign rights to a legal entity for global licensing.

Human creative input must be documented to ensure copyright/patent enforceability.

Use trade secrets for core AI models to avoid human authorship issues.

Draft flexible, jurisdiction-specific licensing agreements.

Secure upstream licenses (training data, third-party IP) before licensing content abroad.

📌 8. Summary Table: IP & Licensing Strategies