Pascal's Chatbot Q&As
Posts
EUIPO Conference on Copyright (and AI): High-quality AI systems must reflect European cultural expressions and values; if we miss this moment, much of Europe’s cultural diversity will be forgotten.

EUIPO Conference on Copyright (and AI): High-quality AI systems must reflect European cultural expressions and values; if we miss this moment, much of Europe’s cultural diversity will be forgotten.

The Munich Regional Court’s GEMA/OpenAI ruling is a key reference point for defining the limits of TDM exceptions, reproduction, and communication to the public.

“Copyright should not only reward past creativity – it must enable the next generation of innovation.”
“High-quality AI systems must reflect European cultural expressions and values; if we miss this moment, much of Europe’s cultural diversity will be forgotten.”
“Literary and artistic creativity for the AI-native generation begins with a prompt.”
(Important for Scholarly Publishers: scientific creativity may also begin with an AI prompt → role of authoritative, licensed scholarly data grows.)
“Freedom for AI trainers to use copyrighted materials—far-reaching freedom—is necessary to build high-quality AI systems… coupled with obligations to ensure cultural diversity is carried into the systems.”
“Technology is advancing rapidly; we will see changes in how AI affects the press, media, and other copyright-dependent sectors.”(Implication: research publishing and knowledge sectors will be next.)

“In the EU we are amongst the pioneers with the AI Act… GPAI models must comply with copyright rules, including e-reservation of rights under the TDM exception.”
“The transparency templates will not be the final answer—Europe must closely monitor impact and build further tools for licensing and enforcement.”
“Content crosses borders instantly, making enforcement, licensing, and fair remuneration more challenging. The framework remains territorial and not yet fully harmonized.”
“EUIPO and the Commission will support the conclusion of licensing agreements between the creative sectors and AI providers.”
“2026 will be the year of copyright in the European Commission—we will reassess the entire 2019 DSM directive, press publishers’ rights, online platform rules, and creators’ remuneration.”(Major signal of upcoming reforms affecting scholarly content, platform liability, TDM, and AI training rules.)

“Licensing is the mega-trend; opt-outs under Article 4 TDM should be seen as an invitation to negotiate licenses.”
“CMOs have in their DNA the ability to license diverse repertoires; memorization and embeddings may require licensing beyond TDM exceptions.”
(Embeddings/reproduction theory → potential basis for licensing models for scientific texts.)
“Large-scale AI licensing deals are proliferating—far more than can be shown on a single readable slide.”
“Relying only on market-driven licensing will not ensure diversity or fair author remuneration; big rightsholders will dominate deals.”
(Implication: academic publishers must form licensing coalitions or collective management structures.)
“AI levy schemes could deliver mandatory remuneration to authors and rightsholders, functioning alongside individual licenses.”
“The AI levy could theoretically be set at a level high enough to let human creators compete—€1,000 per prompt or €1,000 per month.”
(Extreme example, but shows political willingness for structural compensation mechanisms.)
“Standardized AI levies create a benchmarking function—authors can compare collective remuneration with individual licensing outcomes.”
“The Commission will explore solutions to ensure creators can benefit from new AI-driven revenue streams.”

“The EU must resolve market fragmentation—this is the single most impactful issue for regulators.”
(Implication: pan-EU licensing for AI training and scholarly content may become policy priority.)
“If we miss the moment to embed European cultural diversity, future AI systems will forget much of our heritage.”
(Risk that LLMs trained mostly on US datasets overshadow European scholarship.)
“Illegal content must not be used for AI training; enforcement actions and lawsuits are increasing.”
“The Munich Regional Court’s GEMA/OpenAI ruling is a key reference point for defining the limits of TDM exceptions, reproduction, and communication to the public.”
(Important precedent for model-weights-as-copies arguments.)
“Digital technologies disrupt licensing, remuneration, and competition—copyright must support innovation without stifling it.”
“Copyright is no longer fit to stand alone—IP rights must operate as a coherent, interconnected system.”
“Exceptions and limitations must evolve to balance copyright with research, cultural heritage, and fundamental rights.”
(Directly relevant to TDM exceptions used by AI labs and scholarly researchers.)

1. Licensing momentum is accelerating

EU institutions expect collective or large-scale licensing for AI training, embeddings, and downstream outputs.

2. Regulatory expansion is certain

The 2019 DSM directive will be reassessed; the TDM regime may be tightened; levies and mandatory remuneration schemes are on the table.

3. EU cultural & knowledge sovereignty is a priority

Europe explicitly wants AI systems built with European culture, values, and data—including scientific knowledge.

4. Scholarly content will be central in the “quality AI” debate

The EU increasingly views quality training data as economic infrastructure. A publisher can position itself as a premium supplier.

5. Future compliance will be complex but creates competitive advantage

AI Act + copyright transparency rules create barriers for non-compliant AI labs. Publishers with well-structured rights data win.