Copyright Challenges In Polish AI-Enhanced Cultural Audio.
š 1. GÅos JarosÅawa Åukomskiego ā Voice Cloning & Copyright / Personal Rights (Poland)
Case Summary
In late 2025, a Warsaw court accepted the first ever Polish lawsuit involving the unauthorised AIāgenerated voice of a real person for commercial use. The case involves JarosÅaw Åukomski, a wellāknown Polish voice actor whose distinctive voice was allegedly used in advertising audio clips generated by AI without his consent, knowledge, or payment.
Legal Issues
Copyright vs. Personality Rights ā Polish copyright law protects creative works, but the use of a real personās voice intersects with protection of personal rights and performersā rights (e.g., the right to control how oneās voice is commercially used).
Voice as Creative Expression ā A human voice used in artistic performance (e.g., narration, acting) is considered part of the copyrighted performance. Generating a synthetic version without licence can violate those rights.
Precedent Potential ā The outcome may shape future disputes involving AI voice cloning ā not only as an issue of moral and personal rights, but also in copyright and related rights contexts.
Why It Matters
If the court finds in favour of Åukomski, this establishes a precedent that AI audio reproducing an identifiable professional voice without consent may violate Polish copyright related rights or personal rights.
š 2. ZAiKS Reservation on Text & Data Mining (TDM) ā Rights Organisations Blocking AI Training
Context
The Polish collective society for authors and composers, ZAiKS, has publicly announced that it opts out of permitting text and data mining (TDM) on works it represents for the purpose of training AI systems. This means AI developers must obtain a licence before using those works in training.
Legal Issues
TDM Exception in EU Law ā The EU Copyright Directive allows TDM without authorisation unless rightsholders opt out. ZAiKS has used that optāout right to effectively restrict AI training on its repertoire.
Licensing Requirements ā Without a licence, training on copyrighted audio (music, speeches, recordings) may constitute infringement.
Cultural Impact ā Collective societies argue AI training threatens livelihoods and revenue streams for artists.
Why It Matters
This is less a court case and more a legal assertion that shapes how AI models can be lawfully trained on Polish copyrighted audio ā and it is actively being enforced that training without permission could be infringement.
š 3. European Copyright Directive Implementation & Copyright Uncertainty
European Framework Impacting Poland
The EUās Directive on Copyright in the Digital Single Market (CDSM) generally provides a limited exception for TDM (text and data mining) but also allows rightsholders to optāout of that exception. Poland, as an EU member state, must implement this into national law.
Legal Issues
Directives vs National Law ā Although the directive is EUāwide, national implementation varies. Polandās lawmakers have debated how to implement exceptions for AI.
Ambiguity on AI Outputs ā Current EU and Polish law do not clearly define when AIāgenerated audio is copyrighted and who owns it.
Ongoing Disputes ā Polish debates over the law are ongoing; technical and legal ambiguity increases enforcement pressure on courts.
Why It Matters
These policy debates arenāt court cases but set the legal framework in which Polish judges will decide disputes over AIāgenerated audio copyright ā including where training copyright material is permitted and where it isnāt.
š 4. European/Comparative Cases Affecting Polish Practice
Because there are few direct Polish court rulings yet on AI copyright, many relevant cases come from other EU jurisdictions but are highly persuasive for Polish judges interpreting EUāaligned copyright law.
A. First European EUIC Court Case on AI Training & Reproduction
A Munich court found that training large language models on lyrics and reproducing protected text from those lyrics can constitute copyright infringement when the AI can reproduce exact copyrighted lines ā even if training was for machine learning purposes. This reasoning likely influences how courts view generative audio.
Key Points:
If an AI memorises and reproduces copyrighted material, infringement is possible.
This undermines claims that training alone should always be āfair useā or permitted under textāandādata mining exceptions.
Although not a Polish ruling, it impacts EU courts interpreting similar matters.
B. DALLāE & Copyright Ownership Cases
EU jurisdictions (e.g., Czech Republic) have ruled that fully AIāgenerated works that lack human authorship are not protected by copyright. This reasoning matters for AIāgenerated audio which lacks human creative input.
Key Issues:
If an AI generates cultural audio without significant human creative input, it may not be eligible for copyright protection.
A human who uses AI as a tool must have significant creative control to claim rights.
C. EU Parliament & Policy Reports
Reports prepared by the European Parliament highlight tensions between generative AI practices and EU copyright principles ā especially around authorship, originality, and transparency. While not court decisions, these reports help shape future legislation and judicial interpretation.
Implications:
Courts may increasingly demand clearer rules on data use, output protection, and remuneration for original creators.
š Common Copyright Challenges in Polish AIāAudio Context
1. Authorship & Originality
Polish law (like most EU law) requires a human creator for copyright protection. Purely AIāgenerated audio tracks may not qualify unless thereās clear human creative input.
ā” This creates legal uncertainty about protection for AIāgenerated cultural audio.
2. Training Data & Infringement
AI models trained on copyrighted audio data (music, speech recordings) without licence risk infringement if they reproduce recognizable portions. Polish TDM implementation and collective society optāouts make this more difficult.
ā” Requires licences or explicit permissions.
3. Voice Cloning & Personal Rights
The Åukomski case shows how copyright and personal rights (related to a voiceās identity, performance, and reputation) intersect. No clear statutory rules yet exist, so courts are setting first precedents.
4. Moral Rights & Attribution
In Poland, moral rights (like the right to be identified as author or performer) are strong. AI use that obscures or misattributes creative input can lead to legal challenges.
5. Fair Use / Dozwolony Użytek
Polish copyright includes a broad concept of fair use (ādozwolony użytekā). However, its application to generative AI is unsettled and subject to interpretive debate ā especially when commercial exploitation is involved.
š§ Summary: Precedents & Trends Impacting AIāAudio in Poland
| Source | Core Issue | Impact on Polish AI Audio |
|---|---|---|
| Åukomski Voice Cloning Case | Unauthorised AI voice use | Sets a domestic precedent on voice rights |
| ZAiKS TDM Reservation | Restricts AI training access | Forces licences before training |
| EU AI Training Cases (Munich) | Training and reproduction infringement | Reinforces need for permission in EU |
| AI Ownership & Authorship Rulings | Only works with human input protected | AIāonly audio may be unprotected |
| EU Policy Reports | Framework for future law | Helps shape courts & lawmakers |
š Key Takeaways
š For AI Developers
ā Ensure you have licence agreements for audio used in training.
ā Clarify human authorship elements if claiming copyright for outputs.
ā Be wary of reproducing identifiable voices or performances.
š For Creators
ā You may have personal and copyright claims if AI uses your voice or audio without consent.
ā Collective societies (like ZAiKS) can influence how AI tools are legally used.
š For Courts
ā Precedents from broader EU cases will influence Polish case outcomes.
ā Judges are actively shaping how Polish copyright law applies to AI audio.

comments