IP Rights In AI Automated Typhoon Induced Landslide Susceptibility Mapping.
1. Introduction
AI-automated landslide susceptibility mapping involves using artificial intelligence, like machine learning models, to predict areas prone to landslides, often triggered by typhoons. The data sources include:
Remote sensing imagery (satellite or drone)
Topographical maps
Meteorological and hydrological data
The AI system processes these datasets and outputs susceptibility maps, sometimes including risk scores per location.
Intellectual property rights come into play in multiple ways:
Copyright – Protects original expression, such as the code of AI models, datasets (if original), or the resulting maps.
Patent – Protects novel AI algorithms or unique methods for processing data for landslide prediction.
Trade Secrets – Protect proprietary methods or data preprocessing techniques.
Database Rights – In some jurisdictions, the structured compilation of data may be protected.
Challenges arise because AI often uses pre-existing datasets, and the output is generated automatically, raising questions about authorship, ownership, and patentability.
2. Key Legal Issues in IP and AI-generated Maps
Authorship – Who owns the rights to AI-generated content? Human programmer, AI system, or institution?
Patentability – Are AI models patentable if they are automated or partly derived from open-source data?
Data Ownership – Using third-party meteorological or topographical data may involve licensing issues.
Liability – If AI-generated maps fail to predict landslides correctly, who is responsible?
3. Case Laws Relevant to AI and IP Rights
While there are no many AI-specific cases for landslide mapping yet, several cases touch on AI-generated works, databases, and patents that are very relevant. I’ll summarize five detailed cases:
Case 1: Naruto v. Slater (2018, Ninth Circuit, USA)
Facts: A macaque named Naruto took selfies using a photographer's camera. The question arose whether the monkey could claim copyright.
Relevance: The court ruled that non-human authors cannot hold copyright.
Implication for AI landslide maps: If AI autonomously generates a susceptibility map without significant human input, copyright may not apply to the AI itself. Human contribution, such as coding the algorithm or interpreting results, becomes critical.
Case 2: Thaler v. Commissioner of Patents (2021, Australia)
Facts: Dr. Stephen Thaler attempted to patent an invention created by his AI system, DABUS, without human intervention.
Outcome: The Australian Federal Court initially rejected the patent but later allowed AI-assisted inventions to be considered if a human inventor contributed creatively.
Implication: AI models for predicting landslides may not be patentable if entirely autonomous. However, human-designed methods, preprocessing steps, or novel AI approaches can be patented.
Case 3: Feist Publications v. Rural Telephone Service (1991, U.S. Supreme Court)
Facts: Rural Telephone Service created a phone directory, which Feist copied. Court ruled that facts themselves are not copyrightable, only original selection or arrangement is protected.
Implication: In landslide mapping, raw elevation or rainfall data cannot be copyrighted. However, the AI-generated map presentation, color-coding, and processed output may qualify for copyright if human creativity is involved.
Case 4: SAS Institute Inc. v. World Programming Ltd (2013, UK & EU)
Facts: World Programming copied functionality of SAS software to run SAS scripts without copying code. SAS argued copyright infringement.
Outcome: Court held that functionality and ideas are not protected, only expression.
Implication: For AI landslide mapping, algorithmic methods themselves (e.g., machine learning model structures) may be patentable, but mere mathematical formulas or model functionalities cannot.
Case 5: Oracle America, Inc. v. Google LLC (2018, USA)
Facts: Google used Java APIs to develop Android. Oracle claimed copyright infringement.
Outcome: Supreme Court held that APIs can be fair use if transformative and functional.
Implication: AI models using open-source datasets or pre-existing models may be legally allowed if their implementation is sufficiently original or transformative.
Case 6: European Court of Justice – Ryanair v. PR Aviation (2014)
Facts: Ryanair had a database of flight schedules; PR Aviation used it without authorization.
Outcome: Court emphasized database rights under EU law for substantial investment, even if individual data points are unoriginal.
Implication: Large meteorological or topographical datasets used in AI landslide prediction may be protected under database rights, especially in Europe.
4. Synthesis and Recommendations
From the above cases, we can extract key principles:
| Aspect | Guidance |
|---|---|
| Authorship | Ensure human creative input in AI mapping to claim copyright. |
| Patentability | Focus on novel AI methods, preprocessing, or hybrid human-AI workflows. |
| Data Use | Obtain licenses for third-party datasets; respect database rights. |
| Liability | Clearly define responsibility for AI-generated predictions in contracts. |
| Open-source AI | Transformative use or integration with proprietary methods is safer legally. |
5. Practical Implications for AI Landslide Mapping
Human-in-the-loop models are essential for copyright protection.
Patent protection is feasible for unique AI algorithms, especially those optimizing typhoon-landslide prediction models.
Database licensing is critical, especially for meteorological data.
Trade secrets can protect preprocessing pipelines or model training strategies.
Liability clauses are necessary if maps are shared with government agencies or private companies.
In short, intellectual property rights in AI-generated typhoon-induced landslide mapping require a layered strategy:
copyright for human-contributed expressions,
patents for inventive methods,
database protection for structured datasets,
trade secrets for internal AI workflows.
RELATED Blog
comments