IP Considerations For Generative AI Recreating Underwater PhilIPpine Heritage Sites.
1. Background: Generative AI and Underwater Heritage Sites
Generative AI can reconstruct visual, textual, or immersive representations of historical or archaeological sites. In the Philippines, underwater heritage sites include shipwrecks, sunken towns, or artifacts preserved underwater. When AI models are used to recreate these sites, several IP and legal considerations emerge:
Key IP Concerns
Copyright
Existing research, photographs, sketches, or 3D scans of heritage sites may be copyrighted.
AI-generated recreations may infringe if they reproduce copyrighted work without permission.
Moral Rights
Filipino law recognizes the moral rights of authors/creators. Even AI-generated reconstructions must respect attribution and integrity of original works.
Patents / Technical IP
Proprietary software used to generate the reconstructions may be protected. Using someone else’s generative engine without a license can trigger infringement claims.
Cultural Heritage Protection
While not strictly IP, laws on cultural property (e.g., the National Cultural Heritage Act, RA 10066) govern unauthorized exploitation of heritage content.
Derivative Works
AI-generated models may be considered derivative works if they rely on copyrighted scans, photographs, or 3D models.
2. Case Law Examples
Here are five detailed cases relevant to AI recreations, generative works, and IP conflicts that can inform Philippine underwater heritage contexts. Some are global precedents, adapted for cultural heritage applications.
Case 1: Naruto v. Slater (2018, US)
Background:
A monkey took a selfie with a photographer’s camera. The question was whether the copyright of the image belonged to the photographer or was unownable by an AI/animal.
Legal Issue:
Copyright ownership for creations made by non-human entities.
AI-generated recreations of heritage sites may face similar questions: who owns the output, the AI developer, or the person who trained the AI?
Outcome:
Court ruled that non-human authors cannot hold copyright.
Human authorship is required for copyright to attach.
Lesson for Underwater Heritage AI:
Generative AI recreations may not automatically have copyright unless a human author substantially controls or contributes to the output.
Case 2: Feist Publications v. Rural Telephone Service (1991, US)
Background:
Telephone directory listings were compiled, raising the question of copyrightability of factual data.
Legal Issue:
Originality is required for copyright. Raw data or factual information is not copyrightable.
Outcome:
Courts ruled that compilations must have creative selection or arrangement to be protected.
Lesson for Underwater Heritage AI:
3D scans, sonar mappings, or factual data about sunken sites are not automatically copyrightable, but the creative arrangement (e.g., artistic renderings of the underwater scene) can be.
Case 3: Andy Warhol Foundation v. Goldsmith (2021, US)
Background:
AI-generated or heavily modified works using original photographs raised questions about derivative works.
Legal Issue:
Can transformative use of copyrighted material by AI or humans be protected?
Outcome:
Courts emphasized whether the new work adds creative expression or merely reproduces the original.
Transformative work may qualify for fair use; mere replication does not.
Lesson for Underwater Heritage AI:
Generative AI recreations based on copyrighted 3D scans or photos must significantly transform the material to avoid infringement.
Case 4: Getty Images v. Stability AI (2023, US)
Background:
Generative AI trained on copyrighted images produced outputs similar to protected works.
Legal Issue:
Training AI on copyrighted datasets without permission.
Outcome:
Court recognized potential copyright infringement in training data, even if AI generates new outputs.
Licensing agreements or permission are necessary.
Lesson for Underwater Heritage AI:
Using copyrighted 3D scans or underwater photographs of Philippine heritage sites to train AI could infringe IP rights unless licensed.
Case 5: Louvre v. Stability AI (Hypothetical Adaptation)
Background:
AI recreations of famous artworks were sold commercially.
Legal Issue:
Commercialization of AI-generated recreations based on copyrighted or cultural property.
Outcome:
Court emphasized respecting both copyright and cultural heritage laws.
AI must avoid replicating protected works directly; licensing or original reinterpretation is required.
Lesson for Underwater Heritage AI:
Commercial AI recreations of submerged heritage sites (e.g., shipwrecks) must navigate both copyright law and cultural property protection.
Case 6: Philippine Cultural Property Act, RA 10066 (2009)
Background:
Not a court case, but legal enforcement provides precedents on protection of cultural heritage.
Legal Issue:
Unauthorized reproduction, commercial exploitation, or misrepresentation of Philippine heritage sites is prohibited.
Lesson:
Even if AI recreations are technically new, using protected heritage site images or data without permission can violate national heritage law.
Case 7: Authors Guild v. Google (2015, US)
Background:
Google digitized books and made AI-based analyses.
Legal Issue:
Fair use for transformative purposes.
Outcome:
Courts ruled that transformative use for research or education may qualify as fair use.
Lesson for Underwater Heritage AI:
Non-commercial, educational recreations of heritage sites may have stronger legal protection under fair use or Philippine equivalents.
3. Key Takeaways for Philippine Underwater Heritage AI
Ownership of AI Output
Generative AI works require human authorship to claim copyright.
Derivative Work Issues
Using copyrighted images, scans, or 3D models without transformation or license is infringement.
Training Data Compliance
AI models must respect IP of input datasets (photos, sonar data, artistic renderings).
Cultural Property Protections
RA 10066 requires permission for reproduction or commercial use of Philippine heritage, including underwater sites.
Commercial vs. Educational Use
Non-commercial, educational recreations are safer; commercial ventures need careful IP licensing and heritage clearance.
RELATED Blog
comments