Copyright OwnershIP In Collaborative AI Research Projects In Poland

06 Mar 2026 --
0 Comments

🔎 Key Legal Framework in Poland (Applicable to AI Research)

Under Polish law (the Act on Copyright and Related Rights of 1994 as amended), copyright protects only works that are the result of creative activity by a human being. This means:

AI alone cannot be an author or rights‑holder — copyright protection applies only if a natural person contributed original creative expression.

In collaborative AI research, ownership depends on human contributions and contractual arrangements (e.g., between academic partners, industry and university consortia).

AI outputs without meaningful human creative input typically do not receive copyright in Poland.

This human‑authorship requirement is consistent with EU copyright principles implemented in Polish law.

📚 Case Law Examples and Legal Precedents

Because AI‑specific copyright disputes in Poland are still developing, many of the “cases” relevant to AI build on analogue or foundational principles from Polish and EU jurisprudence.

1. 🇵🇱 Case: Sąd Najwyższy, I CSK 156/15 (2015) — Audiobook Derivative Rights

Facts: A publisher released an audiobook without the author’s permission.

Holding: The Polish Supreme Court held that:

An audiobook is a derivative work — even if the text is read aloud, it involves reproduction of the original work.

Permission from the author was required.

Relevance to AI Research:

In collaborative AI projects, if AI is used to create derivative content (e.g., summaries or outputs based on existing copyrighted texts), rights holders must authorize use.

AI research outputs that reproduce protected material could be treated as derivative — triggering copyright obligations.

Key Legal Principle: Copyright protects derivative works; recreating or distributing them without consent may be infringement.

2. 🇵🇱 Case: Sąd Najwyższy, I CSK 352/12 (2013) — Moral Rights and Alterations

Facts: A publisher altered a novel to fit audiobook timing without the author’s approval.

Holding: The Supreme Court reaffirmed that:

Authors’ moral rights (such as integrity of the work) are protected.

Unauthorized modification violated the author’s rights.

Implication for AI Collaboration:

Even in AI‑assisted research, humans remain morally responsible for outcomes.

If AI alters or processes copyrighted material, research collaborators must respect original creators’ moral rights.

Key Legal Principle: Human creative and moral inputs are strictly protected and cannot be overridden by automated processes.

3. 🇵🇱 Case: Sąd Apelacyjny w Warszawie, I ACa 54/16 (2016) — Collective Rights and Licensing

Facts: A company created an audiobook using music from a copyrighted soundtrack without licensing.

Holding: The appellate court held that:

Rights managed by collective societies (like ZAiKS in Poland) require appropriate licences.

Using unlicensed content is infringement.

Relevance to AI Research:

Collaborative AI research often uses datasets drawn from copyrighted works.

Without licence agreements (particularly when using content managed by rights organisations like ZAiKS), research tools and AI models may infringe.

Key Legal Principle: Collective rights organisations can control licensing even within research contexts.

4. 🏛️ EU/ECJ Principle: ECJ, C‑5/08 Infopaq (2009)

Facts: A company used 11‑word excerpts of newspapers in a database.

Holding: Copying even small excerpts can infringe if they contain original expression.

Implication for Poland:

Polish courts generally follow EU interpretations of copyright.

AI systems that copy or reproduce original expressions may be infringing even if only fragments are used.

In collaborative AI research, training data must be licensed or fall within lawful exceptions.

Key Legal Principle: Substantially original expression is protected, no matter how short.

5. 🏛️ EU/ECJ Principle: ECJ, C‑145/10 Painer (2011)

Facts: Photography case emphasising the originality requirement.

Holding: Copyright protects works that show personal intellectual creation (human contribution).

AI Research Context:

Polish law interprets originality similarly: where human creative choices are identifiable, copyright may subsist.

AI outputs with meaningful human editorial control (e.g., in research reports) can generate copyright for human collaborators.

Key Legal Principle: Human artistic or intellectual contributions underpin copyright eligibility.

6. 📊 ZAiKS Reservation on TDM Activities (No Court Decision Yet)

Not a court judgment, but a significant legal position by ZAiKS (Polish collective rights organisation) has case‑law potential:

ZAiKS has formally stated that Text and Data Mining (TDM) — including machine learning and generative AI training involving works it represents — is infringement unless licensed.

This stance challenges the EU Directive TDM exception — especially where Polish implementation allows rightsholders to opt out.

Implication: In collaborative AI research using copyrighted datasets, rights holders could enforce licensing unless TDM exceptions are firmly established.

7. ✍️ Administrative Position on AI Authorship in Poland

Although no Polish court has yet ruled directly on AI‑generated content, published decisions elsewhere in the EU showing AI‑heavy work may not be copyrighted without human creative input influence Polish interpretation:

A European case (AI gallery dispute) found that AI alone cannot be an author — only humans with decisive creative control can claim copyright.

In Polish Collaborative Projects:

Collaborative AI outputs (e.g., articles, data analysis, models) may only attract copyright if clear human choices were embedded in creation, such as manual curation, editing, direction of research methodology, or interpretation.

🧠 Practical Legal Principles for Collaborative AI Research in Poland

✅ 1. Human Creators Must Be Identified

Works must show meaningful human authorship beyond machine output to be copyrighted.

Research reports, summaries, AI‑assisted papers, and analyses can be treated as copyright works if human creative decisions are documented.

✅ 2. Contracts Are Essential

Ownership and rights in AI collaborations are primarily governed by contracts between research partners, sponsors, universities, and developers.

Without contracts specifying ownership, economic and moral rights may be unclear.

✅ 3. Licensing for Training Data

AI research models trained on copyrighted works must respect licensing rules — rights holders (including collective societies like ZAiKS) may require permissions.

✅ 4. Derivative and Moral Rights Must Be Respected

Outputs derived from existing works (e.g., adaptations, summaries) require consent and attribution.

✅ 5. AI Tools and Algorithms

Purely algorithmic code is protected if it truly embodies original creative choices by humans; otherwise, trade‑secret and contract protections are more reliable.

📜 Summary Table: Cases & Their Significance

Case / Legal Source	Jurisdiction	Key Point for AI Research	What It Teaches
I CSK 156/15 (Audiobook)	Poland	Derivative works need permission	AI outputs that reproduce copyrighted material must be licensed.
I CSK 352/12 (Moral rights)	Poland	Protects integrity of original works	Human creators maintain moral rights even in AI outputs.
I ACa 54/16 (Licensing via ZAiKS)	Poland	Use of collective rights	Research must secure proper rights for datasets.
ECJ Infopaq	EU	Small excerpts can infringe	AI training & extraction may require consent.
ECJ Painer	EU	Human creative contribution matters	Human involvement is author basis.
ZAiKS TDM Reservation	Poland (non-judicial)	TDM without licence may be infringement	AI model training must respect rights unless exception applies.
EU AI Authorship Precedent	Europe	AI not an author without human creative input	AI research outputs alone are not copyrighted.

📍 Final Takeaways

AI alone cannot own copyright in Poland — human creators must be identifiable for works to be copyrightable.

Collaborative research outputs (reports, datasets, tools) must be managed through clear contractual ownership agreements between partners.

Licensing and moral rights remain strong — derivative works and training data use must be authorized, especially where content is protected.

Emerging disputes (like ZAiKS’s TDM stance) are shaping future Polish AI copyright enforcement.