Ownership Allocation In AI-Generated Cross-Border Innovations.
1. Introduction: Ownership Challenges in AI-Generated Cross-Border Innovations
AI-generated innovations—whether software algorithms, inventions, digital art, or scientific outputs—pose complex ownership issues, especially when developed, deployed, or used across borders.
Key challenges include:
Determining the inventor/author: Can AI itself hold ownership, or must a human be named?
Patent and copyright allocation: Different jurisdictions recognize AI inventorship differently.
Licensing and assignment: How rights transfer when multiple countries are involved.
Corporate ownership: Assigning rights from AI systems or contractors to a company.
Cross-border enforcement: Differing rules on AI-generated works complicate IP litigation.
Ownership allocation must consider:
Patent law
Copyright law
Trade secrets
Contracts and licensing
Data usage rights
2. Legal Principles Governing AI Ownership
2.1 Inventorship under Patent Law
Many jurisdictions (US, EU) require a human inventor.
Some jurisdictions (Australia) recognize AI as an inventor if AI made substantive contributions.
Corporate ownership usually arises via assignment agreements from inventors.
2.2 Authorship under Copyright Law
AI-generated works may require sufficient human contribution to qualify for copyright.
Ownership is often assigned to employers or commissioning entities.
2.3 Contractual Allocation
Employment agreements or licensing contracts determine who owns outputs generated by AI.
Cross-border projects require consistent assignment clauses to avoid disputes.
2.4 Data and Input Ownership
Training data often contains third-party content; ownership of AI outputs may be contingent on legal use of input data.
3. Key Case Laws on AI Ownership
Here are six detailed cases relevant to AI-generated cross-border innovation ownership.
Case 1: Thaler v. Commissioner of Patents (Australia, 2021)
Facts:
Stephen Thaler listed an AI system, DABUS, as the inventor of two patent applications filed in Australia.
Legal Issue:
Can AI be recognized as a legal inventor?
Court Ruling:
Federal Court of Australia held that AI can be named as an inventor.
Ownership can then be assigned to a corporation.
Recognized that AI may qualify if it made an inventive contribution.
Cross-Border Implication:
Companies filing patents internationally must adjust inventor names per jurisdiction.
In Australia: AI can be listed.
In US and EU: a human must be listed, even if AI generated the invention.
Ownership Strategy:
Assign AI-generated inventions to the corporate entity via agreements.
Maintain records showing AI’s contribution to support filings in jurisdictions recognizing AI inventorship.
Case 2: Thaler v. USPTO (United States, 2021)
Facts:
The same DABUS patent applications were filed in the US.
Legal Issue:
Can AI be listed as an inventor under US patent law?
Court Ruling:
US Federal Circuit held AI cannot be listed as an inventor.
Only natural persons are recognized as inventors.
Cross-Border Implication:
For US filings, companies must identify human operators or developers as inventors.
AI-generated contributions are assigned to companies via agreements to ensure ownership.
Case 3: Authors Guild v. Google (US, 2015)
Facts:
Google scanned millions of books for AI-powered search. Authors sued for copyright infringement.
Court Ruling:
Court held Google’s use was transformative fair use.
Training AI does not infringe if output is non-substitutive.
Ownership Implication:
AI training data ownership does not automatically transfer output rights.
Companies must ensure data usage agreements and maintain proper licensing for cross-border compliance.
Case 4: Warhol Foundation v. Goldsmith (US Supreme Court, 2023)
Facts:
Warhol created art derived from Lynn Goldsmith’s photographs. Goldsmith claimed copyright infringement.
Court Ruling:
Commercial licensing without consent violated copyright.
Transformative defense was limited.
Ownership Implication:
For AI-generated works based on third-party material, ownership is contingent on permissions.
Cross-border projects must consider local copyright law before assigning ownership.
Case 5: European Patent Office Decision on AI Inventorship (EPO, 2022)
Facts:
Applications listing AI as inventors were rejected.
Legal Issue:
Can AI be listed as an inventor in Europe?
Ruling:
EPO rejected AI inventorship.
Assignments must designate human inventors, even if AI contributed substantially.
Cross-Border Implication:
Multi-jurisdiction filing strategies must:
Assign ownership to corporations.
Designate humans in EU filings.
Document AI contribution for internal records.
Case 6: Oracle America v. Google (Java API, US, 2021)
Facts:
Google reused Java APIs in Android OS. Oracle claimed copyright infringement.
Court Ruling:
Supreme Court ruled that Google’s use was fair use because it was transformative and necessary for software innovation.
Ownership Implication:
AI-generated software using existing frameworks must consider licensing and derivative ownership.
Ownership can be retained by AI developers if transformative use and compliance with license terms are demonstrated.
4. Ownership Allocation Strategies for Cross-Border AI Innovation
A. Patent Assignments
File corporate-owned patents.
Adjust inventor names by jurisdiction (AI in Australia, humans in US/EU).
Use PCT filings to protect inventions globally.
B. Copyright and Moral Rights
Determine whether AI-generated works qualify for copyright in each country.
Assign ownership via employment or commissioning agreements.
Explicitly address moral rights in jurisdictions like EU where they cannot be waived easily.
C. Trade Secrets
Protect AI models, training datasets, and proprietary code through confidentiality agreements.
Cross-border collaborations must comply with export control regulations and secrecy laws.
D. Contracts and Licensing
Use international IP contracts to clarify ownership.
Include clauses for AI contributions, derivative works, licensing, and commercialization rights.
Specify jurisdiction for dispute resolution and applicable law.
E. Data Ownership
Ensure training datasets are licensed for AI development.
Document dataset provenance and cross-border permissions.
Avoid infringement when transferring AI models internationally.
5. Key Takeaways from Case Law
| Case | Jurisdiction | Legal Principle | Ownership Implication |
|---|---|---|---|
| Thaler v. Commissioner (DABUS) | Australia | AI can be inventor | AI contributions can be assigned to company |
| Thaler v. USPTO | USA | Only humans recognized | Human must be listed; assign corporate ownership |
| Authors Guild v. Google | USA | Fair use for AI training | Training data rights do not transfer output ownership automatically |
| Warhol v. Goldsmith | USA | Transformative use limited | AI outputs using third-party material require permission |
| EPO AI Inventor Decision | Europe | AI not recognized | Human inventors must be designated for patent filings |
| Oracle v. Google | USA | Transformative software use | Ownership can be retained if fair use or license compliance is ensured |
6. Practical Guidelines for Cross-Border AI Ownership
Document contributions of AI and humans.
Assign AI-generated inventions to corporations via contracts.
Adjust filings per jurisdiction (AI inventor vs human inventor).
Secure licenses for datasets and third-party content.
Use trade secrets to protect proprietary models and code.
Ensure cross-border contractual clarity for ownership, licensing, and commercialization.
Integrate regulatory compliance with ownership allocation to avoid disputes.
7. Conclusion
Ownership allocation in AI-generated cross-border innovations requires a multi-layered strategy:
Patent law: adjust inventor designations and corporate assignments.
Copyright law: clarify human involvement and third-party permissions.
Trade secrets: protect models and datasets internationally.
Contracts: explicitly assign rights across jurisdictions.
Case law demonstrates:
AI can be inventor in some countries (Australia), but not in US/EU.
Human inventorship and corporate assignments are essential for global protection.
Copyright and licensing are key for derivative AI works.
Cross-border strategies must reconcile legal, regulatory, and technological realities.
RELATED Blog
comments