IPR Challenges In Licensing AI-Driven Earthquake Prediction And Response Systems.
I. Overview: What Makes AI‑Driven Earthquake Prediction & Response Systems IPR‑Sensitive?
AI systems for earthquake prediction and response typically include:
AI Algorithms – Machine learning models trained on seismic data.
Data Sets – Seismic monitoring data, historical earthquake records.
Software Platforms – Tools that visualize predictions.
User Interfaces – Dashboards delivered via licenses.
Outputs & Predictions – Forecasts, risk maps, alerts.
Combined Services – APIs, cloud access, analytics.
These raise unique IPR concerns:
Who owns the data used to train the AI?
Who owns the AI models?
Can the output of an AI model be copyrighted or patented?
How do licenses allocate rights to use, distribute, modify, or commercialize the system?
What about rights of third parties whose data is embedded?
What happens when multiple stakeholders co‑develop the technology?
II. Four Major IPR Challenges in Licensing AI Earthquake Systems
1. Copyright & Authorship of AI Outputs
AI systems create predictions, maps, alerts, and reports. Do these outputs have copyright? And if so, who owns them?
2. Patentability of AI Methods
Can the underlying AI algorithms, seismic feature‑extraction techniques, or novel prediction systems be patented?
3. Data Rights & Database Protection
Earthquake prediction depends on massive seismic datasets from sensors and institutions. Who owns the right to use this data?
4. Licensing & Innovation Incentives
Balancing open science (often needed for public safety) with private investment (commercial sustainability) creates tension in licensing agreements.
III. Relevant Case Laws & Why They Matter
Below are seven key cases whose principles shape IPR licensing for AI‑based earthquake systems:
1️⃣ Naruto v. Slater (2018)
Jurisdiction: U.S. Ninth Circuit
Facts
A macaque monkey took photographs with a photographer’s camera. Litigation arose over who owned the copyright.
Holding
Non‑humans cannot own copyright.
Why This Matters
AI systems generating earthquake predictions raise a similar question:
Is an AI the “author” or “creator”?
If the AI model generates a prediction or map, can anyone claim copyright?
Principle Established
Copyright only vests in humans (or legal persons), not machines.
📌 Implication for Earthquake AI Licensing
Licenses must explicitly state that:
Humans (developers, institutions) or entities own the copyright in AI outputs.
AI systems cannot themselves be treated as rights‑holders.
2️⃣ Feist Publications v. Rural Telephone (1991)
Jurisdiction: U.S. Supreme Court
Facts
Feist used phone listings from Rural Telephone Service. Rural alleged infringement of its directory.
Holding
Facts and mere compilations are not copyrightable unless there’s originality.
Why This Matters
AI‑driven prediction systems often compile raw seismic data, sensor readings, timestamps, etc. Just arranging data may not satisfy copyright.
Principle Established
Creativity matters. Raw data is not protected unless selected or arranged with creativity.
📌 Implication
Seismic datasets, maps, or prediction tables with no added creativity may not be copyrightable.
Licenses must clarify whether only software/code is protected — not the underlying facts.
3️⃣ Thaler v. USPTO (2023)
Jurisdiction: U.S. Federal Court
Facts
Stephen Thaler attempted to list an AI system as the “inventor” on a patent application.
Holding
Only humans can be inventors under U.S. law.
Why This Matters
AI earthquake algorithms may be technically novel. But:
Patents must list human inventors.
AI cannot legally be an inventor.
Principle Established
AI may contribute but cannot be recognized as a legal inventor.
📌 Implication for Licensing
Patents on AI methods must assign inventorship to people.
Licensing must carefully document contributions when AI was used in creation.
4️⃣ Bridgeman Art Library v. Corel Corp. (1999)
Jurisdiction: U.S. Southern District of New York
Facts
Bridgeman claimed copyright on exact reproductions of public domain artwork.
Holding
Exact reproductions of public domain works lack originality.
Why This Matters
Earthquake data and maps derived directly from science may similarly be factual rather than creatively original.
Principle Established
Simply reproducing public domain facts doesn’t create a new copyright.
📌 Implication
Output of AI forecasting based only on factual sensor data may not be protected.
Licensing must focus on software and proprietary processes, not raw outputs.
5️⃣ Sega v. Accolade (1992)
Jurisdiction: U.S. Ninth Circuit
Facts
Accolade reverse engineered Sega software to make compatible games.
Holding
Reverse engineering for interoperability is sometimes allowed — but only if it doesn’t copy protected expression.
Why This Matters
AI earthquake systems often need to interoperate with:
Government databases
Sensor networks
Emergency response systems
Licenses may limit interoperability.
Principle Established
Reverse engineering to achieve interoperability can be defensible — unless it infringes protected expression.
📌 Implication
Licenses must clearly define interoperability rights.
AI model access to external systems must be negotiated.
6️⃣ Oracle v. Google (2014)
Jurisdiction: U.S. Federal Courts
Facts
Google used Java API structures in Android. Oracle sued for copyright infringement.
Holding
APIs may have some copyright protection based on structure and organization.
Why This Matters
AI systems expose APIs for earthquake forecasting and response.
Principle Established
APIs are not entirely free to copy — structure matters.
📌 Implication
Licensing must explicitly define rights to use APIs.
Users cannot assume free access without terms.
7️⃣ Alice Corp. v. CLS Bank (2014)
Jurisdiction: U.S. Supreme Court
Facts
Alice Corp. claimed a patent on a computerized method for mitigating financial risk.
Holding
Abstract ideas implemented with computers are not always patentable.
Why This Matters
AI seismology patents may be challenged as “abstract ideas” — e.g., using data + computer may be insufficient for patentability.
Principle Established
To be patentable, inventions must be more than abstract ideas; they need inventive concept.
📌 Implication
Earthquake AI patents must show a technical improvement.
Licensing revenue may depend on strong patents.
IV. Practical Licensing Challenges Based on These Cases
1. Who Owns the Output?
Due to Naruto and Feist:
AI predictions and maps may not automatically be protected unless shaped by human creativity.
Licensing agreements must clearly assign ownership of outputs.
👉 For example:
“All AI‑generated prediction data shall be owned by the Licensee, subject to Licensee’s compliance with training data rights.”
2. Patent Ownership & Inventorship
Because of Thaler and Alice:
Patents on algorithms must list humans.
Licenses should clarify patent rights and ownership.
👉 Agreements often include:
Joint ownership clauses
Assignment of rights from contributors
3. Data Licensing & Database Rights
Earthquake prediction depends on seismic data from:
Public agencies
Private sensors
Academic institutions
Feist teaches that data as raw facts may be unprotected unless the data structure or compilation adds creativity.
👉 Licensing must address:
Data access rights
Data privacy and security
Third‑party data restrictions
4. Interoperability & API Licensing
From Sega and Oracle v. Google:
APIs and interfaces may have protective rights.
Licenses must define whether reverse engineering or interoperability is permitted.
5. Moral & Ethical Dimensions
Unlike art or literature, earthquake AI systems impact human safety:
Errors can cause harm.
Licensing must address liability, risk, and misuse rights.
This is not directly an IPR case, but is part of licensing risk management.
V. Summary of Key Legal Lessons
| Challenge | Relevant Case | Key Legal Principle |
|---|---|---|
| Authorship of AI output | Naruto v. Slater | Only humans (or legal persons) can own copyrights |
| Copyrightability of data | Feist | Mere facts/compilations lack originality |
| Patent authorship | Thaler v. USPTO | AI cannot be an inventor |
| Originality of reproductions | Bridgeman | Exact reproductions of public domain lack copyright |
| Interoperability | Sega | Reverse engineering may be allowed for interoperability |
| API ownership | Oracle v. Google | Structure of APIs can be protected |
| Patentability test | Alice | Abstract ideas are non‑patentable without inventive concept |
VI. Key Takeaways for Licensing Agreements
To license AI‑driven earthquake systems:
Define Ownership
Outputs
Models
Improvements
Protect Patents
Claim technical inventions
Assign inventorship properly
Clarify Data Use
Get rights for training datasets
Respect third‑party restrictions
API Rights
State clear rights for integration and interoperability
Liability Terms
Allocate risk for misuse, errors, and predictions
RELATED Blog
comments