Comparative Study Of AI-Related Patent Litigation Precedents In France And Germany.
I. INTRODUCTION
Artificial Intelligence has created complex legal questions in patent law, particularly regarding:
Inventorship – Can an AI system be an inventor?
Patentability – Are AI-generated inventions patentable?
Inventive Step & Technical Character – How do courts assess AI-based inventions?
Use of copyrighted data for AI training – How does this affect innovation and patent enforcement?
France and Germany, as EPC (European Patent Convention) member states, generally align with European Patent Office (EPO) standards. However, Germany has produced more explicit judicial reasoning on AI inventorship, while France’s AI litigation is still emerging.
II. GERMANY: DETAILED CASE ANALYSIS
Germany has been more active in judicial reasoning concerning AI and patent law.
1. Federal Court of Justice (Bundesgerichtshof – BGH), X ZB 5/22 (DABUS Case, 2024)
Background
An applicant filed a German patent application listing an AI system called DABUS as the sole inventor. The German Patent and Trade Mark Office refused the application because no natural person was named.
The case eventually reached the Federal Court of Justice (BGH), Germany’s highest civil court.
Legal Issue
Can an AI system be named as an inventor under German Patent Law (Patentgesetz)?
Court’s Reasoning
The Court held:
The term “inventor” under German Patent Law refers to a natural person.
Inventorship is linked to:
Legal capacity
Moral rights
Transfer of rights
An AI system lacks legal personality and cannot:
Hold rights
Transfer rights
Exercise moral rights
The Court emphasized that patent law is structured around human creative activity.
Holding
AI cannot be named as inventor.
However, the Court clarified that:
If AI is used as a tool,
The human who controls or directs the inventive process
must be named as inventor.
Significance
This is the most authoritative German precedent on AI inventorship. It firmly establishes:
Human inventorship remains mandatory.
AI-assisted inventions are patentable, but only if a human inventor is identified.
Companies must carefully document human contribution when AI systems are involved.
2. Federal Patent Court (Bundespatentgericht), 11 W (pat) 5/21 (2021)
Background
This was an earlier stage of the DABUS litigation. The applicant attempted to name the AI as inventor and argued that modern interpretation should accommodate AI-generated inventions.
Legal Question
Can “inventor” be interpreted dynamically to include AI systems?
Court’s Analysis
The Court rejected a dynamic interpretation and stated:
Patent law presupposes human creativity.
Legislative reform would be required to allow AI inventorship.
Courts cannot reinterpret statutory language to include non-human entities.
Interestingly, the Court allowed a procedural possibility:
A human applicant could state that the invention was generated using AI,
But the inventor must still be a natural person.
Significance
This case laid the groundwork for the 2024 BGH decision. It reinforced the traditional anthropocentric structure of patent law.
3. AI and Technical Character – German Case Law on Software/Algorithm Patents
While not exclusively about AI, German courts have addressed the patentability of computer-implemented inventions that are highly relevant to AI technologies.
Federal Court of Justice – “Logikverifikation” and related cases
German jurisprudence consistently holds:
Algorithms “as such” are not patentable.
A computer-implemented method must show technical character.
The invention must solve a technical problem using technical means.
Relevance to AI
AI-based inventions often rely on:
Mathematical models
Neural networks
Training algorithms
German courts examine whether:
The AI model produces a technical effect beyond abstract data processing.
The system controls physical processes (e.g., robotics, medical devices).
If AI merely processes business data, patent protection is unlikely.
If AI improves image processing in medical diagnostics, patentability is stronger.
Significance
Germany applies strict scrutiny to inventive step in AI patents. Technical contribution must be clear and measurable.
4. Kneschke v. LAION e.V. (Hamburg Regional Court 2024; Higher Regional Court 2025)
Though primarily a copyright case, it significantly affects AI innovation and patent ecosystems.
Facts
A photographer sued LAION, a non-profit organization that created large AI training datasets including copyrighted images.
The claim alleged unauthorized reproduction for AI training.
Legal Issue
Does creating AI training datasets infringe copyright, or is it protected under Text and Data Mining (TDM) exceptions?
Court’s Reasoning
The Hamburg Court held:
AI training qualifies as text and data mining.
Under German copyright law implementing the EU Digital Single Market Directive:
Reproduction for scientific research can be lawful.
LAION’s activities were non-commercial and research-oriented.
The Higher Regional Court upheld dismissal of the claim.
Significance
This case impacts patent litigation indirectly:
It protects dataset creation for AI development.
It supports AI research environments.
It strengthens Germany’s position as AI-friendly in research contexts.
III. FRANCE: DETAILED CASE ANALYSIS
France has fewer finalized AI patent cases but remains aligned with EPC doctrine.
5. French DABUS Applications (INPI Practice and Judicial Position)
Background
Applications similar to the German DABUS filings were submitted in various jurisdictions.
The French Patent Office (INPI) rejected AI-only inventorship.
Legal Position
Under the French Intellectual Property Code:
The inventor must be a natural person.
Moral rights of inventorship are inalienable.
Only a person can exercise these rights.
Although no landmark French Supreme Court ruling exists yet, administrative practice mirrors German and EPO reasoning.
Significance
France adheres strictly to:
Human inventorship
Traditional moral-rights-based conception of invention
6. SNE, SNAC, SGDL v. Meta (Paris Judicial Court, 2025 – Pending)
Facts
French publishing and authors’ associations sued Meta alleging that its generative AI model was trained on copyrighted French works without authorization.
Legal Issues
Unauthorized reproduction for AI training
Economic parasitism
Violation of authors’ rights
Relevance to Patent Ecosystem
While not a patent case, it affects:
AI innovation frameworks
Investment in AI research
Data accessibility for AI development
If courts restrict training data use, AI innovation and related patent filings may decline.
If courts allow broad TDM exceptions, innovation may accelerate.
Significance
France may adopt a stricter stance than Germany regarding commercial AI training.
7. French Case Law on Computer-Implemented Inventions
French courts, following EPC standards, require:
A technical problem
A technical solution
A non-obvious contribution
AI-related patents face scrutiny under inventive step analysis.
French courts have invalidated high-tech patents where:
The contribution was deemed algorithmic abstraction.
The “technical effect” was insufficient.
Thus, AI patents in France must demonstrate concrete technical impact (e.g., signal processing improvement, hardware optimization).
IV. COMPARATIVE ANALYSIS: FRANCE vs GERMANY
| Issue | Germany | France |
|---|---|---|
| AI as Inventor | Explicitly rejected by Federal Court of Justice (2024) | Rejected by administrative and EPC-based interpretation |
| Judicial Depth | Extensive reasoning at highest court level | No Supreme Court ruling yet |
| AI as Tool | Accepted if human directs invention | Same approach |
| Patentability of AI Algorithms | Strict technical-effect requirement | Strict technical-effect requirement |
| Training Data Litigation | More permissive (LAION case) | Possibly stricter (Meta case pending) |
| Innovation Climate | Strong support for AI research exceptions | More protective of authors’ rights |
V. KEY DOCTRINAL DIFFERENCES
1. Judicial Activism
Germany has issued clearer doctrinal reasoning via the BGH.
France relies more heavily on EPC harmonization and administrative practice.
2. Copyright vs Innovation Balance
Germany appears more innovation-friendly in dataset creation.
France may emphasize authors’ economic rights more strongly.
3. Inventorship Philosophy
Both systems remain fundamentally anthropocentric.
Neither recognizes machine autonomy as legally creative.
VI. CONCLUSION
From the examined cases:
AI cannot be an inventor in either France or Germany.
Human contribution remains central to patent validity.
AI-assisted inventions are patentable if:
Technical character exists,
Inventive step is demonstrated,
A human inventor is identified.
Germany has produced clearer Supreme Court jurisprudence.
France’s major AI-related litigation is currently emerging, especially in copyright-training disputes.
Overall, Germany has taken a more structured and innovation-accommodating judicial approach, while France remains cautious and protective of traditional intellectual property principles.
RELATED Blog
comments