Patent Law Implications For AI-Created Pharmaceutical Formulas.

01 Apr 2026 --
0 Comments

1. Background: AI in Pharmaceutical Invention

AI is increasingly used in drug discovery, predicting molecular structures, optimizing chemical synthesis, or identifying drug candidates. While AI speeds up R&D, it raises unique questions under patent law:

Inventorship: Traditional patent law requires a human inventor. Can AI be listed as an inventor?
Patentability: Are AI-generated inventions “novel” and “non-obvious” if AI algorithms are used?
Ownership & Rights: Who owns patents generated with AI—the programmer, AI operator, or institution?

2. Patentability Criteria & AI

Pharmaceutical formulas must meet:

Novelty (35 U.S.C §102 in US law): Must not be publicly disclosed.
Non-obviousness (35 U.S.C §103): Must not be obvious to someone skilled in the field.
Enablement (35 U.S.C §112): Must allow a skilled person to make/use the invention.

AI challenges these because it can generate highly novel compounds, but the human contribution may be minimal.

3. Key Case Laws

Case 1: DABUS AI Inventorship Cases (US & UK)

Facts:
The AI system DABUS created inventions, including a “food container” and “neural flame device.” Its creators applied for patents, listing the AI as inventor.

Decisions:

USPTO (2021): Rejected applications, stating only humans can be inventors.
UKIPO (2021): Same conclusion—AI cannot be listed as an inventor under UK law.

Implications for pharma:

AI-generated drug formulas cannot name the AI as inventor.
Human oversight or intervention is necessary for patent filing.
Ownership defaults to the human who directed or created the AI.

Takeaway: AI is a tool, not a legal inventor.

Case 2: Thaler v Comptroller General of Patents (UK Supreme Court, 2022)

Facts: Chris Thaler argued that DABUS should be recognized as inventor.

Decision: UK Supreme Court upheld the previous ruling that patent law requires a human inventor.

Key Point:

This reinforced that AI-generated pharmaceuticals need a human inventor for legal protection.
In pharma R&D, AI can guide design, but patents must list a human who contributed intellectually.

Case 3: European Patent Office (EPO) – DABUS Applications

Facts: EPO considered whether AI could be an inventor for European patents.

Decision:

EPO rejected applications citing Article 81 EPC—inventor must be a natural person.
No recognition of AI as a legal inventor.

Implications:

EU pharma patent filings must involve human contribution.
AI-generated compounds may still be patentable if a human researcher contributed creatively.

Case 4: In re Kubin (Fed. Cir. 2009)

Facts: Though pre-AI, In re Kubin is foundational for biotech/pharma. It addressed obviousness of DNA sequences for therapeutic use.

Decision:

Patents were denied because sequences were considered “obvious” if standard molecular techniques could generate them.

Relevance to AI:

AI can rapidly generate many sequences. If AI uses standard algorithms, the resulting formula may be considered obvious, affecting patentability.
Novelty and inventive step remain crucial.

Case 5: Amgen Inc. v. Sanofi (US Supreme Court, 2023)

Facts: Patent disputes over biologics (antibody drugs) and the interpretation of enablement in biotech patents.

Decision:

The Court stressed patents must enable a skilled person to reproduce the invention without undue experimentation.

Implications for AI:

AI-generated drug formulas must be fully disclosed in patent applications.
Simply providing an AI output may not suffice if a human skilled in pharma cannot reproduce it.

Case 6: Ariosa Diagnostics v. Sequenom (2015, US Fed. Cir.)

Facts: Patents on non-invasive prenatal testing (NIPT) methods were challenged.

Decision:

Court invalidated patents as they claimed natural phenomena without inventive steps.

Relevance:

AI in pharma may discover natural compounds or pathways.
Patents must show human-guided invention or innovative application, not mere AI discovery of natural facts.

4. Legal Takeaways for AI-Generated Pharmaceuticals

Inventorship:
Only humans can legally be inventors. AI is a tool, not an inventor.
Ownership:
Whoever controls or programs the AI and applies human creativity usually owns the patent.
Novelty & Obviousness:
AI output may be obvious unless human researchers contribute creative selection or combination.
Enablement & Disclosure:
Full disclosure is critical; AI formulas must be reproducible by skilled humans.
International Variations:
- US & UK: AI cannot be inventor.
- EPO: Same; AI can assist but human inventor required.
- Some jurisdictions are exploring AI as co-inventor, but no binding precedent yet.

5. Practical Steps for Pharma Companies

Always include human researchers in patent applications.
Document human decision-making in AI-assisted drug design.
Evaluate obviousness carefully—AI might generate trivial variants.
Provide detailed enablement sections—how a human can reproduce AI results.

Summary Table of Cases

Case	Jurisdiction	AI Relevance	Key Outcome
DABUS	US/UK	AI inventorship	Only humans can be inventors
Thaler v. UKIPO	UK	AI inventorship	Reinforced human-only inventorship
EPO DABUS	EU	AI inventorship	Rejected AI inventor, human required
In re Kubin	US	Obviousness in biotech	AI output must show inventive step
Amgen v. Sanofi	US	Enablement in pharma	Full disclosure required for reproducibility
Ariosa v. Sequenom	US	Natural phenomenon	AI cannot patent discoveries of nature alone

In short, AI is revolutionizing pharmaceutical discovery, but under current patent law:

AI cannot be a legal inventor.
Human involvement is essential for ownership, inventorship, and meeting legal standards.
Patent strategies must emphasize human creative input, reproducibility, and non-obviousness.