EU AI Code Adds Copyright and Training-Data Transparency Duties
The EU General-Purpose AI Code of Practice includes copyright-policy and training-content transparency measures for participating model providers.
Topics: arts and humanities, science and technology, economics
Example
What happened: The EU General-Purpose AI Code of Practice includes copyright-policy and training-content transparency measures for participating model providers.
Use it for: as AI learns from human creative work, societies must decide whether artists should know, consent to and be paid for that use.
Why it matters: generative AI models are trained on huge amounts of text, images and music made by people; requiring developers to have a copyright policy and disclose training content gives creators a first step towards control.
Limit: A voluntary code can clarify compliance but participation, enforcement, and fair compensation remain contested.
Key facts
- The EU General-Purpose AI Code of Practice includes copyright-policy and training-content transparency measures for participating model providers.
How to use this example in a GP essay
How should societies balance generative AI innovation with creators' rights?
Claim
as AI learns from human creative work, societies must decide whether artists should know, consent to and be paid for that use
How the evidence supports it
generative AI models are trained on huge amounts of text, images and music made by people; requiring developers to have a copyright policy and disclose training content gives creators a first step towards control
Limitation
A voluntary code can clarify compliance but participation, enforcement, and fair compensation remain contested.
Relevance
It gives a current rule for questions on AI and the arts, especially whether technology threatens creative livelihoods.
Limitations
- A voluntary code can clarify compliance but participation, enforcement, and fair compensation remain contested.
Evaluations
arts and humanities evaluation
Support
Transparency about training content lets artists find out whether their work was used, which is the first step to asking for consent or payment.
Counterargument
The code is voluntary and does not itself pay artists, so creators may gain information but no income.
Rebuttal
Information is the basis for negotiation and legal claims; without knowing what was used, artists could not bargain at all.
Additional support
The code asks participating providers to adopt copyright policies, which treats respect for creators as part of responsible AI.
science and technology evaluation
Support
Clear expectations on copyright let AI developers build with less legal uncertainty, which can support innovation as well as protect creators.
Counterargument
Disclosure duties may slow smaller developers who lack legal teams, giving an edge to large firms.
Rebuttal
The burden is real but modest compared with lawsuits; a shared code can lower costs for everyone by setting one standard.
Additional support
Because the code sits under the EU AI Act framework, it shows regulators using guidance to fill gaps before formal rules apply.
economics evaluation
Support
If training on creative work becomes visible, a market for licensing that work can develop, giving artists a new income source.
Counterargument
Licensing deals may favour large publishers and labels, leaving independent artists with little.
Rebuttal
Who benefits from licensing depends on how collective rights groups are organised, a question the code opens rather than settles.
Additional support
The case shows a conflict between two growth sectors, AI and the creative industries, that governments must balance.
Sources
- European Commission: General-Purpose AI Code of Practice