Patent Filing For AI-Generated Algorithms In Finance And Analytics
Patent Filing for AI-Generated Algorithms in Finance and Analytics
Artificial Intelligence (AI) has become a powerful tool in finance and analytics. AI algorithms are now used for risk assessment, fraud detection, high-frequency trading, portfolio optimization, and predictive analytics. Many of these algorithms are autonomously generated or optimized by AI, raising complex legal questions regarding patentability, inventorship, and ownership.
I. Legal Framework for Patenting AI Algorithms in Finance
Patent Law Basics
A patent protects inventions that are novel, non-obvious, and useful.
In finance, AI algorithms may be software-based methods for:
Predictive modeling of market trends
Fraud detection using neural networks
Portfolio optimization via reinforcement learning
Challenges
Many jurisdictions, including the U.S. and EU, exclude abstract ideas from patentability.
Pure mathematical algorithms or business methods may not be patentable unless they produce a technical effect or solve a technological problem.
Inventorship and Ownership
If AI autonomously generates an algorithm, who is the inventor?
Courts generally require human inventorship.
Ownership often follows:
Employer (work-for-hire)
Contractual agreements
Joint ownership for collaborative AI projects
Licensing
Licensing AI-generated algorithms involves IP rights, data usage, and regulatory compliance.
Financial sector algorithms also face restrictions from data privacy regulations (e.g., GDPR, CCPA).
II. Key Case Laws Relevant to AI-Generated Algorithms in Finance
1. Thaler v. Vidal (2022, USA)
Facts: Stephen Thaler filed patent applications for inventions created by AI system DABUS, including algorithmic devices.
Issue: Can AI be recognized as an inventor in the U.S.?
Decision: Rejected; inventors must be natural persons.
Relevance: AI-generated trading algorithms or predictive models cannot list AI as inventor. Human programmers or strategists must be identified.
2. Thaler v. Comptroller-General of Patents (2023, UK Supreme Court)
Facts: UK patent applications filed for AI-invented devices, including algorithmic systems.
Decision: AI cannot be an inventor under UK patent law.
Implication for Finance: AI-generated financial algorithms must have human inventors to qualify for patent protection.
3. Thaler v. Commissioner of Patents (2021, Australia)
Facts: AI was named as inventor in patents for various inventions, including algorithm-based devices.
Decision: Initially allowed by Federal Court, but overturned by Full Federal Court.
Takeaway: Reinforces global consensus—human inventorship is mandatory.
4. Alice Corp. v. CLS Bank International (2014, USA)
Facts: Concerned patent eligibility of computer-implemented business methods.
Decision: Introduced two-step test for abstract ideas:
Determine if claim is directed to an abstract idea.
If so, check if the claim contains an “inventive concept” sufficient to transform the idea into patent-eligible invention.
Relevance to Finance: AI algorithms for trading, risk analysis, or fraud detection must do more than implement an abstract idea on a computer. Must provide concrete technological improvement, such as speed, efficiency, or system integration.
5. DDR Holdings v. Hotels.com (2014, USA)
Facts: Concerned patent eligibility of internet-based business methods.
Decision: Patents are valid if solving a technological problem rather than a purely abstract or economic problem.
Relevance to Finance:
AI algorithms for fraud detection or analytics may be patentable if they solve a technical problem, such as processing data in real-time or optimizing computational efficiency.
6. Diamond v. Chakrabarty (1980, USA)
Facts: Patent granted for genetically engineered bacteria.
Decision: “Anything under the sun made by man” is patentable if novel and useful.
Implication: Human-guided AI-generated algorithms in finance can be patented, provided human contribution exists.
7. University of Utah v. Max-Planck-Gesellschaft (2013, USA)
Facts: Dispute over inventorship in genetic research.
Decision: Inventorship requires human contribution to conception.
Relevance: If AI autonomously optimizes a predictive trading model, the human designer who sets goals and parameters is considered the inventor.
8. Naruto v. Slater (2018, USA)
Facts: Copyright claim for a selfie taken by a monkey.
Decision: Non-human entities cannot hold IP.
Relevance: Reinforces principle that AI cannot be an inventor or owner.
III. Patent Filing Strategy for AI-Generated Finance Algorithms
Identify Human Inventors
Determine who contributed to:
Defining the problem
Designing the AI workflow
Selecting datasets
Interpreting results
Demonstrate Technical Improvement
For software/algorithm patents, show technical advantages:
Faster execution
Reduced computational resources
Enhanced data security
Document AI Contribution
Maintain logs of AI outputs and human decision points.
Crucial for establishing inventorship and avoiding legal disputes.
Draft Claims Carefully
Focus on method, system, and device claims.
Emphasize practical applications in finance (e.g., fraud detection, automated trading, predictive modeling).
Licensing
Define ownership of algorithms, models, and datasets.
Address joint ownership if multiple parties contribute AI expertise.
Ensure compliance with financial regulations and data privacy laws.
IV. Practical Implications
| Scenario | Patent Filing Outcome |
|---|---|
| AI autonomously generates a trading strategy with human guidance | Human(s) who set parameters and interpret output can be named as inventors |
| AI independently creates predictive analytics algorithm | No patent without human contribution |
| Collaborative development across banks and AI vendor | Joint ownership; licensing agreement essential |
| AI optimizes risk assessment model for efficiency | Patentable if technical improvements demonstrated (Alice Corp test) |
V. Emerging Challenges
Human Inventorship
Hard to determine for fully autonomous AI.
Abstract Idea Limitation
Many AI algorithms in finance are considered abstract mathematical methods.
Data Dependency
Patent eligibility may depend on unique training datasets, which must comply with privacy regulations.
Global Differences
US, EU, UK, and Australia currently require human inventors; patent offices differ in approach to algorithm-based inventions.
Conclusion
AI cannot be a legal inventor, but human contributors can patent AI-generated algorithms if they provide conceptual input.
Patentability requires:
Novelty, non-obviousness, and industrial applicability
Technical improvement beyond abstract idea (especially post-Alice Corp)
Key Cases:
Thaler v. Vidal (USA)
Thaler v. Comptroller-General (UK)
Thaler v. Commissioner of Patents (Australia)
Alice Corp v. CLS Bank (USA)
DDR Holdings v. Hotels.com (USA)
Diamond v. Chakrabarty (USA)
University of Utah v. Max-Planck (USA)
Naruto v. Slater (USA)
Best Practices:
Document all human involvement in AI algorithm design.
Emphasize technical innovation and system improvements.
Use robust contracts to clarify ownership and licensing rights.
RELATED Blog
comments