Oscar Guinea / Sep 2026
Image: Shutterstock
On 1 September, the US Department of Justice (DoJ) entered the copyright fight between The New York Times, OpenAI and Microsoft. The DoJ asked a Manhattan court to treat training large language models as fair use (a US rule that can allow use of copyrighted material without a licence when it serves a new purpose). It warned that a more restrictive approach would slow scientific progress, reduce prosperity and weaken national security. The Trump administration believes that copyright now belongs in the same conversation as geo-economic rivalry and defence.
Europe has approached the question from a narrower starting point: creators against large technology companies. The European Parliament’s March resolution calls the current rules inadequate for generative AI, seeks an additional licensing framework, and demands stronger control for people and businesses that own creative works, including authors, publishers and performers. Meanwhile, the EU Court of Justice is examining a case about whether a chatbot can infringe a publisher’s rights when its answer potentially reproduces part of a news article. The case, Like Company v Google Ireland, remains pending after a March hearing. These questions require legal answers. But they cannot settle a policy that reaches well beyond a dispute between a publisher and a platform.
Artificial intelligence is a general-purpose technology. It will be used in factories, financial services, medicine, logistics and the creative industries. For writers, musicians and illustrators, AI brings opportunities but also risks: AI can compete in markets that have traditionally relied on human creativity. But policymakers cannot weigh that risk in isolation. It must also account for the opportunities that AI creates across the wider economy. The scale matters. The EU’s creative industries produced €202 billion in gross value added in 2023. Yet the combined value added of ICT, finance, automotive, pharmaceuticals and chemicals was more than nine times larger. In other words, a mere 1 per cent fall in those sectors would require the EU’s creative industries to expand output by more than 9 per cent to offset it.
Europe should not sacrifice its creative industries on the altar of competitiveness. But nor should it damage the larger economy by misdiagnosing the problem. The creative sector has a serious concern: loss of work and income as algorithms enter markets that have traditionally rewarded human creative labour. That is labour displacement. It has happened before, from the theatre stage confronting television to the shop floor confronting automation. It calls for labour-market policies and, where justified, support for the arts. It does not follow that access to the material from which AI learns should be restricted as a means of compensating the creative sector for these potential losses.
The cost of getting copyright policy wrong would be spread across the EU economy in the form of slower innovation. But the costs would not be shared equally: the burden would fall most heavily on smaller firms, which lack the capital, legal expertise and negotiating power to navigate complex licensing arrangements. The largest US firms can absorb licensing fees and employ armies of lawyers. A European start-up trying to adapt a model to help airlines cut fuel use through better route and maintenance planning cannot. The perverse result would be to entrench the very foreign incumbents that Europe says it wants to challenge, while making its own companies dependent on imported models.
Europe still has an opening in AI, though not primarily in frontier models. Its comparative advantage lies in diffusing and adapting AI across its industrial base: tailoring existing models to manufacturing, health, or finance. For this, firms need lawful access to their most precious material: data. Article 4 of the Copyright in the Digital Single Market Directive allows text and data mining where the material is legally accessible and its owner has not expressly reserved that use. It is one of the few EU digital rules that makes this key digital input more abundant rather than costlier.
The EU should retain that balance: punish actual copying but resist a compulsory licensing regime that turns every act of computational learning into a negotiation. This is not a plea to ignore creators. It is a reminder that copyright policy now shapes Europe’s industrial strength, economic resilience and security. The Americans have noticed. Europe should, too.
This article draws on ECIPE’s policy brief, The Copyright Trap, which is available to read and download on ECIPE’s website.
