GPT-4o: AIPPI’s resolution stands out as one of the most balanced, comprehensive, and forward-looking legal proposals to date on AI and copyright.

AIPPI’s AI & Copyright Resolution – Implications, Value, and Recommendations for AI Stakeholders

by ChatGPT-4o

What is AIPPI and What Does it Do?

The International Association for the Protection of Intellectual Property (AIPPI) is a leading global non-profit organization dedicated to the development and improvement of intellectual property (IP) laws. With members in over 125 countries, AIPPI brings together legal experts, practitioners, and academics to study and propose harmonized solutions to IP challenges across jurisdictions.

AIPPI’s most impactful mechanism is its World Congress, during which national and regional groups submit legal analyses on study questions. These are distilled into resolutions that express AIPPI’s official stance and recommendations on pressing IP issues. These resolutions are non-binding but are highly influential—informing lawmakers, courts, and international treaty bodies such as WIPO, WTO, and national legislatures.

In 2025, at its World Congress in Yokohama, AIPPI adopted a resolution on one of the most contested and unresolved legal topics in the digital age: the intersection of Artificial Intelligence (AI) and Copyright.

Overview of the 2025 AIPPI Resolution on AI & Copyright

The resolution focuses on whether and how copyright laws should apply to:

1. The use of copyrighted works to train AI systems,

2. The outputs generated by those systems, and

3. Whether the AI systems themselves can be considered infringing articles.

It draws from 39 detailed national reports and represents a globally harmonized yet nuanced position, advocating a balanced approach between technological innovation and copyright protection. Among its key proposals are:

• Harmonization is desirable across jurisdictions to clarify the legality of AI training and output use (Point 1).

• Prior authorisation from rightsholders should be required to train AI on copyrighted materials, unless an exception applies (Point 2).

• Limited exceptions should apply for non-commercial, public interest purposes such as education and scientific research (Point 4).

• In commercial use cases, opt-out rights and financial compensation mechanisms should be implemented (Point 5).

• Any exceptions must comply with the Berne Convention’s three-step test to prevent abuse (Point 6).

• Transparency obligations for AI developers and providers are essential to enable rights enforcement (Point 7).

• Outputs should not be automatically infringing just because they are in the same style as a copyrighted work (Point 10).

• AI systems can be considered infringing if they were intentionally trained with unlawfully obtained content or designed to generate infringing outputs (Point 13).

• Multiple parties can be held liable for infringing outputs, including developers, exploiters, and deliberate prompters (Point 14).

• The resolution outlines a range of sanctions and remedies, from damages to destruction of infringing systems and outputs (Points 15–19).

This resolution was co-drafted with input from the Dutch study group, including legal experts like Noa Naaman, Clemens Molle, Anke Strijbos, Jasper Klopper, Meyke Rietveld, Frodo Ferro, and chaired by Thijs van Aerde—highlighting the Netherlands’ active role in shaping global IP standards in the AI age.

Added Value of This Resolution

The AIPPI resolution delivers significant added value in several ways:

1. Clarity and Structure in a Fragmented Landscape: AI-related copyright disputes are escalating worldwide, but legal consensus remains elusive. AIPPI’s resolution brings structured legal reasoning, aiming to reduce uncertainty and regulatory fragmentation across jurisdictions.

2. Bridging Stakeholder Interests: The resolution mediates between AI developers’ need for access to training data and copyright holders’ demand for protection and remuneration. The opt-out and compensation provisions (Point 5) represent a pragmatic compromise.

3. Global Harmonization Blueprint: While non-binding, the resolution acts as a normative template for lawmakers, courts, and trade negotiators seeking balanced and globally consistent rules.

4. Preserving Creativity and Cultural Integrity: Through attention to moral rights(Point 9) and the distinction between style vs. content replication (Point 10), it helps preserve artistic reputation without stifling innovation.

5. Strong Enforcement Recommendations: The remedies section (Points 15–19) strengthens the enforcement toolbox and deterrence capacity for rightsholders—critical as LLMs and generative AI proliferate globally.

6. Futureproofing: The resolution considers adaptiveness of AI after deploymentand indirectly anticipates regulatory challenges around retraining, model updating, and user-prompt liability (Point 14c).

In sum, the resolution demonstrates that responsible AI governance is not incompatible with robust copyright enforcement—a critical message for both creators and tech innovators.

Recommendations for AI Makers

Based on AIPPI’s resolution, AI developers and vendors should:

1. Implement Training Data Transparency Mechanisms
   Ensure that the origin, type, and copyright status of training data can be traced and disclosed. Metadata embedding, dataset audits, and public registries could aid compliance.

2. Respect Opt-Out Mechanisms and Prepare for Licensing Models
   Support TDM opt-out frameworks (e.g. EU TDM opt-out registry) and prepare systems to comply with pay-per-use or compensation obligations for commercial applications.

3. Segment Use Cases
   Distinguish clearly between non-commercial/public interest training and commercial deployments. Limit model reuse across those boundaries to avoid legal risk.

4. Avoid Style-Based Filtering Alone
   Simply filtering out copyrighted “styles” may not be sufficient to avoid infringement claims—invest in more robust content filtering and detection technologies.

5. Avoid Infringing Design Goals
   Ensure AI models are not deliberately fine-tuned to replicate or exploit specific copyrighted works unless authorized—especially in fields like media, gaming, and education.

6. Prepare for Liability Along the Value Chain
   Build indemnity clauses, rights tracking systems, and risk-sharing mechanisms into contracts with downstream users and distributors of your AI systems.

7. Engage with Policymakers
   Proactively contribute to national and international legislative efforts using AIPPI’s resolution as a conversation starter.

Do I Agree with AIPPI’s Position?

Yes, largely. AIPPI’s resolution stands out as one of the most balanced, comprehensive, and forward-looking legal proposals to date on AI and copyright.

It does not reject AI development or large-scale training practices outright, but places reasonable legal guardrails and fairness mechanisms around their use. The emphasis on prior authorization for commercial use and the opt-out + compensation model reflects the reality that AI’s value is often built on creative labor that must not go uncompensated.

Some points could benefit from clearer technical implementation guidance—e.g., how to enforce point 7(a) in the context of opaque foundation models—but this is arguably outside the scope of a legal resolution.

In conclusion, AIPPI’s resolution offers a pragmatic and principled foundation for resolving tensions between copyright law and generative AI. It should be taken seriously by developers, legislators, and rights holders alike.