Patent Disputes In Ai Inventions
PATENT DISPUTES IN AI INVENTIONS
1. Introduction
Artificial Intelligence (AI) inventions present unique challenges for patent law due to:
Questions of inventorship: AI as a co-inventor?
Patent eligibility: Algorithms, software, and abstract ideas
Ownership issues: Employer vs. AI developer vs. AI itself
Prior art and obviousness: Machines generating prior art automatically
These disputes are emerging globally, including in India, USA, Europe, and UK.
2. Legal Challenges in AI Patent Disputes
2.1 Inventorship
Traditional patent law requires a natural person as an inventor.
AI raises questions:
Can an AI system be an inventor?
How to assign ownership if AI generates the invention autonomously?
2.2 Patent Eligibility
Many AI inventions rely on mathematical models or algorithms, which can fall under Section 3(k) of the Indian Patents Act, 1970 or 35 U.S.C §101 (abstract ideas) in the USA.
The challenge: distinguishing between non-patentable software and patentable AI-implemented inventions.
2.3 Obviousness & Prior Art
AI systems can autonomously generate inventions.
Determining novelty and inventive step can be complicated if AI-generated content exists in prior art.
3. Key Case Laws on AI and Patent Disputes
CASE 1: Thaler v. USPTO (DABUS Case, USA/UK/EU)
Facts:
Stephen Thaler attempted to register patents with DABUS, an AI system, listed as the inventor.
Applications included:
AI-designed beverage container
AI-generated neural network-based invention
Issues:
Can AI be recognized as an inventor?
Who owns the patent if AI is the inventor?
Decisions:
USPTO (2021): Rejected; only natural persons can be inventors.
UKIPO (2022): Confirmed that AI cannot be an inventor under UK law.
EPO (European Patent Office): Also rejected the application.
Principles:
Inventorship requires human intervention.
AI-generated inventions may need a human applicant to claim ownership.
CASE 2: Alice Corp v. CLS Bank (USA, 2014)
Facts:
Alice Corp patented a computer-implemented method for reducing settlement risk.
CLS Bank argued it was abstract and non-patentable.
Issue:
Is an AI-implemented software patent eligible under U.S. law?
Decision:
The Supreme Court held the patent invalid, as it covered abstract ideas merely implemented on a computer.
Principles:
Software/AI patents are closely scrutinized for abstractness.
AI inventions need technical implementation beyond mere algorithms.
CASE 3: IBM AI Invention Dispute (USA, 2020)
Facts:
IBM AI created a new semiconductor design.
The dispute arose between IBM engineers and AI-generated work: who owns the patent?
Issue:
Inventorship and contribution assessment.
Outcome:
Courts recognized human supervision and intervention as critical.
Patent granted to IBM as applicant, citing human-directed AI use.
Principle:
AI can assist invention, but human inventorship is required for legal recognition.
CASE 4: European Patent Office – DABUS Appeal (EP 2022)
Facts:
Similar to the Thaler case. DABUS listed as the sole inventor.
Issue:
Can AI itself be listed as inventor?
Decision:
EPO refused; only humans are allowed as inventors under the European Patent Convention (EPC).
Principle:
Confirms global consensus: AI is a tool, not a legal inventor.
CASE 5: DeepMind AlphaFold Patent Disputes (UK/EU, 2021–2022)
Facts:
DeepMind developed AlphaFold, an AI system for predicting protein structures.
Multiple entities sought patents for AI-generated methods.
Issue:
Whether AI predictions and outputs can form patentable inventions.
Decisions:
Patents were granted only where human involvement in method design was demonstrated.
Pure AI predictions without human inventive input were rejected.
Principle:
AI results can contribute to patentable inventions if humans exercise creative control.
CASE 6: Indian Patent Office – AI-Generated Inventions (2023 Draft Guidelines)
Facts:
Indian Patent Office addressed AI-generated inventions in consultation.
Principles:
Inventor must be a human (draft aligns with global trends).
AI may assist in prior art search, optimization, or testing, but cannot be inventor.
Applications must highlight human contribution clearly.
CASE 7: Starkey Labs AI Hearing Aid Invention Dispute (USA, 2022)
Facts:
AI optimized hearing aid circuitry.
Dispute arose over ownership vs. AI contribution.
Decision:
Patents granted to company employing humans who guided AI.
Courts emphasized that human inventive concept must exist.
4. Emerging Trends
AI as a tool, not inventor – universally recognized.
Human supervision is critical – patents are granted only if human contribution can be demonstrated.
AI can accelerate prior art searches and optimization but cannot autonomously own IP.
Global harmonization underway – US, UK, EU, and India all emphasize human inventorship.
Dispute focus shifting to ownership and contribution – who gets credit for AI-assisted inventions.
5. Practical Guidance for AI Invention Patents
Clearly document human involvement in invention.
Describe AI contribution as a tool or assistant.
Avoid claiming AI as inventor.
Ensure technical implementation beyond abstract AI algorithms.
Consider drafting collaborative invention clauses for AI-assisted teams.
6. Conclusion
Patent disputes in AI inventions revolve around:
Inventorship – AI cannot be inventor.
Patentability – Only technical, non-abstract implementations are patentable.
Ownership – Assigned to human inventors or their organizations.
Global consensus – USPTO, EPO, UKIPO, and India align on human inventorship.
AI is a powerful tool for invention, but IP law currently protects only humans as inventors. Courts are likely to continue scrutinizing AI-assisted patents for human contribution, novelty, and inventive step.
RELATED Blog
comments