Copyright Governance Of UkrAInian AI-Based Music Restoration Projects.

25 Feb 2026 --
0 Comments

📜 1. Ukrainian Copyright Law and AI: Legal Framework

Core Principles

Under the Law of Ukraine “On Copyright and Related Rights” (effective 1 Jan 2023):

Only a natural person can be an author. A work must result from the creative labour of a human to attract full copyright protection.

AI alone is not recognized as an author. Any output from AI without significant human creative input cannot be copyrighted as a traditional “work.”

Non‑original AI outputs are protected only under a special “sui generis” regime (Article 33): rights are limited to economic rights and last 25 years. Moral rights (e.g., right to be named as author) do not arise.

This legal context shapes how AI‑based music restoration is governed in Ukraine.

📌 Case‑Style Example 1: The “AI Music Re‑Erasure” Dispute

Scenario: A tech studio uses an AI tool to restore damaged historical Ukrainian folk recordings. The AI processes the raw audio, fills missing parts, and generates a polished music file.

Issue:

Who owns copyright in the restored track?

Analysis:

If the restoration involved substantial human creative decisions (selection of styles, editorial choices, instrumentation choices), the resulting restored music may qualify for regular copyright — because a person added original contributions.

If the AI merely filled gaps without significant human direction, under Ukrainian law it would be a non‑original output generated by a computer program.

In that case, the studio (or the rights holder of the tool/trainer) obtains sui generis rights, limited to economic exploitation (licensing, distribution) for 25 years.

Takeaway:

The line between “copyright” vs “sui generis rights” hinges on human creative involvement, not on how advanced the AI is.

📌 Case‑Style Example 2: “Copyright Claims on Restored Vintage Songs” (Hypothetical)

Scenario: A music historian uploads AI‑restored versions of 1950s Ukrainian songs to a global streaming platform without permission.

Conflict:

Original rights holders (or their heirs) claim infringement, arguing:

The core composition remains their protected work.

Even though the audio has been “improved,” the core musical work is still their property.

Likely Judicial Reasoning:

The original musical work’s copyright remains fully intact (because the original author’s rights persist for 70 years after death or works’ publication).

AI restoration does not extinguish original rights — it modifies or enhances the recording.

Uploading the restored audio without permission still constitutes distribution and public performance of the copyrighted work unless covered by an exception or licensed.

Outcome (Plausible):

The court would rule in favor of the original rights holders, holding the historian liable for infringement (damages, takedown). Only the additional elements contributed by a human editor (e.g., creative arrangements) might attract separate copyright.

📌 Case‑Style Example 3: Ukraine’s First AI‑Sui Generis Registration “Music‑Image Works”

Context from the Ukrainian IP Office:
Ukraine’s IP Office has granted copyright registration for composite works using AI‑generated images when the human’s creative compilation was striking.

Analogous Music Example (Constructed for Explanation):

Suppose a creator uses AI to generate instrumental loops and then arranges them with traditional instruments and songwriting.

Legal Reasoning:

The arrangement and human composition choices are protected by traditional copyright.

The pure AI loops are protected only under sui generis rights (economic rights — e.g., for licensing), not full moral rights.

The end musical work thus has dual protection: traditional rights for the human’s contributions + sui generis rights for AI‑generated elements.

Significance:

This kind of decision shows Ukraine’s willingness to protect hybrid works but clearly separates what is copyright (human effort) from what is sui generis (AI alone).

📌 Case‑Style Example 4: Judicial Policy on AI Outputs in Court (Real)

In 2025 the High Administrative Court of Ukraine ruled that submitting arguments to a court that are solely generated by AI without human reasoning constitutes procedural abuse.

Translating This to Music Copyright:

If a music restorer tries to justify exclusive rights by submitting AI‑only generated legal arguments in court (e.g., “the AI owns the music”), the court will reject it as lacking human legal reasoning.

Similarly, permit decisions (e.g., licensing for use) have to demonstrate human creative intent and analysis.

Principle:

Human accountability and judgment remain central — even in disputes involving AI outputs.

📌 Case‑Style Example 5: Hypothetical — “AI‑Training Dataset Lawsuit”

Scenario: A Ukrainian startup uses a generative AI trained on hundreds of copyrighted tracks to restore and normalize audio. A coalition of Ukrainian composers sues, alleging unauthorized use of their works for training violated rights.

Issues to be Resolved:

Is AI training on copyrighted music without permission an infringement?

Ukrainian law currently does not explicitly allow AI training as a lawful use.

Courts may analogize from reproduction rights: copying for training equals an infringement if done without consent.

Do the restored outputs infringe original works if the AI model learned from them?

This can become a derivative use dispute: if the restored outputs “resemble” protected compositions, the original authors may argue a derivative infringement.

Likely Result (Based on Legal Reasoning):

The court would analyze:

Whether the training stage made copies of original music.

Whether outputs are substantially similar to original works.

If yes, the startup may owe compensation (damages, licensing fees).

This mirrors global litigation trends (labels suing AI‑music companies), but applied under Ukrainian copyright norms.

🧠 Key Governance Themes in AI‑Music Projects

1. AI != Author

AI can assist but cannot be author under Ukrainian law — copyright always attaches to the human contribution.

2. Sui Generis Rights

Ukraine uniquely grants limited economic rights to pure AI outputs, but those are narrower than full copyright.

3. Original Works Remain Protected

Restoring or enhancing a pre‑existing work does not negate the original rights holder’s copyright.

4. Transparency in AI Use

Disclosures identifying what part was AI generated help courts and rights holders assess ownership and infringement risk.

📘 Practical Takeaways for AI‑Music Projects in Ukraine

Issue	Legal Status in Ukraine
AI as author	❌ not allowed
AI output copyright	sui generis (economic rights, 25 years)
Human‑enriched AI output	✔ full copyright for human contribution
Training AI on music without permission	⚠ legally uncertain but potentially infringing
Using AI‑restored music commercially	✔ if rights cleared

🧠 Conclusion

Ukrainian copyright governance around AI‑based music restoration is emerging and nuanced. It draws a clear distinction between:

Human creativity (copyright);

Computer generation (sui generis rights);

Original works and rights holders (still fully protected).

Judicial practice so far emphasizes human reasoning and accountability, and future disputes (especially training data cases) will further shape this fast‑evolving legal area.