Ownership Of AI-Generated Genomic Data In Cross-Border Research.
I. INTRODUCTION
AI-generated genomic data refers to sequences, predictions, or annotations produced by AI systems applied to genomic research. These datasets can arise from:
Predictive modeling (e.g., predicting disease risk from DNA)
Novel variant discovery
Synthetic genomes or gene-editing outcomes
Ownership of such data is complex due to overlapping legal regimes:
Intellectual Property Law – copyright, patent, database rights
Contract Law – agreements between research institutions
Data Protection and Privacy Law – human genomic data is personal data
International Law – cross-border sharing introduces jurisdictional issues
The key legal question is:
Who owns AI-generated genomic data when multiple parties, jurisdictions, and AI systems are involved?
II. LEGAL FRAMEWORK
1. Intellectual Property Considerations
Patentability: AI-generated inventions can be patentable if human inventorship can be claimed (Thaler case principles applied).
Copyright: Raw data itself is usually not copyrightable, but compilations or AI-processed datasets may be.
Database Rights (EU): Sui generis database rights protect investment in obtaining, verifying, or presenting data.
2. Data Protection Laws
EU GDPR: Genomic data is sensitive personal data. Ownership is not granted to AI but to data controllers, with obligations to protect individuals’ rights.
Other jurisdictions: HIPAA (US), Indian Personal Data Protection Act – regulate cross-border transfer of genomic data.
3. Contractual Agreements
Ownership and rights often depend on:
Research collaboration agreements
Licensing of AI software
Terms of AI data generation
III. MAJOR CASES RELEVANT TO AI-GENERATED GENOMIC DATA
Here are more than five important cases that help understand ownership, intellectual property, and AI involvement:
1. Thaler v. Perlmutter (2023) – U.S. District Court
Facts:
Stephen Thaler created AI (“Creativity Machine”) that generated works claimed as authorship. Though this was artistic, the reasoning applies to genomic inventions.
Issue:
Can AI itself be an inventor/owner?
Judgment:
AI cannot be listed as an inventor.
Human who conceives, programs, or uses AI for creative purpose may claim ownership.
Relevance to Genomic AI:
The person or institution operating AI for genomic research is the likely legal owner.
Autonomous AI-generated sequences cannot be patented or copyrighted without human attribution.
2. Naruto v. Slater (2018) – U.S. Ninth Circuit
Facts:
A monkey took photographs; PETA claimed copyright.
Judgment:
Non-human entities cannot hold copyright.
Relevance:
AI, like animals, is not a legal person.
Any AI-generated genomic datasets must have a human or institutional owner.
3. Myriad Genetics, Inc. v. Association for Molecular Pathology (2013) – U.S. Supreme Court
Facts:
Myriad claimed patents on isolated BRCA1 and BRCA2 genes.
Judgment:
Naturally occurring DNA sequences cannot be patented.
cDNA (synthetic DNA) is patentable.
Relevance to AI-generated genomic data:
AI-generated sequences that mimic natural sequences may not be patentable.
Synthetic sequences created by AI with human oversight may qualify for patent protection.
4. Association for Molecular Pathology v. Myriad Genetics (2012) – U.S. District Court / Federal Circuit
Issue:
Clarified patentable subject matter for genetic sequences.
Judgment:
Confirmed distinction between naturally occurring and artificially created DNA.
AI-generated predictive sequences must demonstrate human-directed inventiveness to qualify for patents.
Relevance:
Cross-border AI research must ensure patent filings align with jurisdictions.
Ownership depends on inventive human contribution, not the AI alone.
5. European Union Database Directive Cases – British Horseracing Board Ltd v. William Hill Organization Ltd (2001)
Facts:
BHB claimed database rights over collected horse racing data; WH used the data.
Issue:
Can investment in compiling data confer ownership?
Judgment:
Database rights protect substantial investment in obtaining and presenting data.
Copying entire database infringes these rights.
Relevance:
AI-generated genomic databases may be protected under EU sui generis database rights.
Ownership is conferred on the institution making the investment, not the AI.
6. Moore v. Regents of the University of California (1990) – California Supreme Court
Facts:
John Moore’s cells were used to develop commercially valuable cell lines.
Issue:
Does a patient retain ownership of biological materials?
Judgment:
No property right in excised cells once donated.
Ownership rests with researchers/institutions under informed consent.
Relevance:
Cross-border genomic AI datasets may involve human genomic data.
Consent and agreements determine ownership and data transfer rights.
7. In re Fisher (2015) – U.S. Patent Office / Court
Facts:
University researchers developed a genetically modified plant using computational tools.
Judgment:
Patent granted to humans directing the invention, not the software.
Relevance:
Reinforces human authorship principle in AI-assisted genomic inventions.
Institutions operating AI may claim ownership.
8. Harvard Mouse Patent Case (Oncomouse) – European Patent Office, 1988
Facts:
Harvard patented genetically modified mice.
Judgment:
Patents allowed if humans direct and create novelty.
Relevance:
AI-generated genomic organisms/sequences must have human oversight for ownership.
Cross-border research must navigate differing patent eligibility standards.
IV. CROSS-BORDER RESEARCH ISSUES
Jurisdictional Differences
U.S.: human-directed invention required for patent/copyright
EU: sui generis database rights protect AI-generated datasets
India: computer-generated work ownership attributed to human initiator
Data Transfer Regulations
GDPR restricts export of genomic data to non-adequate countries.
Consent must explicitly cover cross-border AI processing.
Ethical Considerations
Transparency: disclose AI involvement
Benefit-sharing: if data is derived from indigenous populations
Privacy: ensure anonymization or pseudonymization
Contractual Ownership Clauses
Collaboration agreements should specify:
Who owns AI outputs
Licensing terms
Publication rights
V. SUMMARY OF PRINCIPLES
| Legal Aspect | Principle |
|---|---|
| AI as Inventor | Not recognized (Thaler, Naruto) |
| Human Oversight | Required for ownership/patent (Myriad, Fisher, Oncomouse) |
| Database Rights | Protect compilation & investment (EU Directive, BHB v. William Hill) |
| Biological Material | Human subjects retain privacy, not property (Moore v. Regents) |
| Cross-Border | Consent & jurisdiction determine ownership & transfer (GDPR, national law) |
Key Takeaways:
AI-generated genomic data cannot own itself.
Ownership vests in humans or institutions directing AI.
Patent protection requires human inventorship; copyright may protect curated datasets.
Ethical and privacy standards are critical in cross-border research.
RELATED Blog
comments