The European Commission has reopened one of the harder questions surrounding Europe’s technology rulebook: whether copyright law needs another round of intervention now that generative AI has moved from experimental systems into commercial products built on enormous quantities of digital content.

A targeted consultation launched on 29 September asks rightsholders, AI companies, intermediaries, researchers, public authorities, consumer groups, and other organisations for evidence across four areas. Alongside the use of copyright-protected works in artificial intelligence, Brussels is examining online piracy of time-sensitive content such as live events, remuneration rights for music performers and producers, and the treatment of copyright in scientific research.

The consultation does not itself change European law, nor has the Commission committed to a particular legislative response. Instead, it is gathering evidence for what it describes as a possible initiative to improve the copyright environment for creativity and innovation, with responses accepted until 3 November.

Although the scope extends well beyond artificial intelligence, generative AI is likely to draw much of the attention because copyright has become one of the principal points of friction between model developers and industries whose material can enter training datasets. The commercial argument is increasingly less about whether models can produce useful outputs and more about how the inputs used to build them are obtained, documented, licensed, or excluded.

AI rules already reach into copyright

Europe is not approaching that question from a blank sheet of paper because the AI Act already imposes copyright-related duties on providers of general-purpose AI models. Providers must maintain a policy for complying with EU copyright law, including recognition of rights reservations, while also publishing a sufficiently detailed summary of material used to train their models.

Those obligations have applied to providers placing general-purpose models on the EU market since August 2025, while the Commission’s enforcement powers over the relevant requirements became applicable in August 2026. A voluntary General-Purpose AI Code of Practice adds a compliance route covering transparency and copyright, including how model providers should identify machine-readable signals used by rightsholders to reserve works from text and data mining.

The new consultation therefore opens a second track. Whereas the AI Act largely asks model providers how they comply with existing European copyright rules, the Commission is now examining whether the underlying copyright environment itself remains adequate as technology and markets change.

That distinction will be important for companies trying to understand their eventual exposure. Model developers can comply with documentation and policy requirements while disputes continue over licensing, enforcement, and the boundaries of text and data mining. Publishers, musicians, image owners, and other rightsholders can likewise adopt technical opt-out mechanisms without resolving broader questions about remuneration or practical enforcement when training takes place across complex data supply chains.

From compliance mechanics to market structure

The Commission’s consultation questionnaire shows how far the discussion has moved beyond a simple argument over scraping. It includes sections on generative AI licensing, enforcement, AI-generated imitations of performers, piracy, and research, placing the technology inside a wider debate about how digital markets allocate economic value.

Any eventual intervention will have consequences beyond the largest foundation-model companies. Enterprise software vendors increasingly embed third-party models into products, while media businesses, research organisations, data providers, and specialist AI developers are creating licensing arrangements of their own. Greater clarity could make those deals easier to structure, although a more prescriptive framework could also raise the cost of developing or adapting models for the European market.

Research institutions face a related tension because access to large bodies of text and data is fundamental to computational research, but scientific publishing and database rights can complicate reuse. The Commission’s decision to include research copyright in the same exercise suggests that it is examining the wider machinery through which knowledge is licensed and analysed rather than treating generative AI as an isolated dispute.

There is also a competition dimension. Large technology groups can absorb licensing costs, legal review, compliance teams, and negotiations with major content owners more easily than smaller model developers. Rules designed to improve compensation and legal certainty may therefore alter the economics of market entry depending on how licences, opt-outs, and enforcement mechanisms are structured.

Brussels has spent much of the past several years building separate legal frameworks around artificial intelligence, digital platforms, data, and online services. Copyright now sits increasingly across those systems because AI development turns cultural and informational material into an input for software products, while the outputs of those products compete in some of the same markets from which the source material originates.

The consultation leaves the eventual policy route open, but it moves the European copyright debate into a more concrete phase. The Commission is now asking not merely how generative AI fits existing rules, but whether those rules still distribute rights, obligations, and economic value in a workable way.