Blog post by Stephen Wyber
On 8 October 2025, the European Commission launched two key AI documents:
These efforts advance its commitment to ensure that the European Union (EU) keeps up with its main competitors in realising the technology’s potential.
The strategies focus on what the EU can do to support the uptake of AI specifically in the science and research sector, and then more broadly across the economy, society, and government.
These are broad documents, touching on many factors, but they help, even if only tentatively, to underline that there is scope for a much more balanced approach to AI regulation that takes account of the interests of the EU as a whole.
AI Research and AI for Research
The AI in Science Strategy establishes RAISE–the Resource for AI Science in Europe– a ‘virtual institute’ bringing together different strands of work intended to facilitate AI research and uptake, turning Europe into an ‘AI continent’.
In doing so, the Strategy sets a goal of developing ‘a distinctly European approach’. This is defined by:
- Avoiding scattered national approaches, through stronger coordination and cross-border working.
- Approaches taken in other parts of the world, and so better able to respond to political priorities around competitiveness and sovereignty.
- Rejecting “Bad’ AI” – i.e. that which is not “ethical, explainable, transparent, accountable, reliable, safe, human-centric and aligned with human rights and societal values”. This aligns also with priorities around research integrity and trust in science.
The strategy (and so RAISE) has four key areas of focus:
- Excellence: The goal is to attract talent and support networks, as well as develop clear guidelines.
- Computational power: This area covers investment in the capacity necessary for training.
- Research funding: This looks at how to coordinate EU funding better and make it responsive to technological change.
- Data: This area is of particular interest to libraries, given their role in supporting researchers to gain access to the information they need for training and making optimal use of it. Much of what is mentioned here simply refers to existing initiatives, such as data spaces, and how these should cooperate with AI factories.
There is also a welcome reference to a commitment to gather further evidence on how to ensure that publicly funded research results are available for AI training. This is timely: with evidence of efforts by publishers and others to restrict AI training and applications in research institutions, it is welcome that the Commission is exploring this.
At the same time, it is noticeable how the strategy talks little about access to research publications. This is despite the fact that the EU’s own commissioned research underlines that uses of AI in science are as much about possibilities to use it to explore literature as they are to work with pure data. This same research makes only cursory reference to intellectual property, and certainly does not claim that it is central to questions of transparency or ethics.
Applying AI More Broadly
The second document, the Apply AI Strategy, looks at the wider economy, and runs through a range of sectors where there is identified scope to make use of AI to improve performance. These range from utilities, such as energy, the public sector (there is even a factsheet about how the Commission will use AI internally), health, and beyond.
Most initiatives here are focused on sector-specific networks and initiatives, but the Strategy does give consideration to the creative and cultural sector as one that could benefit. At the same time, it is direct about one of the challenges to uptake: strong opposition to (some aspects of) AI can be a barrier.
The response proposed is considerably more balanced than the arguments often heard in the culture debate. Rather than simply seeking to ban all training unless there is payment, the Commission proposes to:
“launch a targeted study to explore the legal challenges related to AI-generated outputs and how cutting-edge technological safeguards and technologies, including AI, could be used to prevent and mitigate the risks of copyright infringing AI content being generated, including by detecting and removing such content.”
In other words, what is on the table is a more proportionate approach that looks at situations that would directly harm creators by reproducing their outputs illicitly.
Finally, a striking aspect of the Apply AI Strategy is the assertion that “AI is rapidly becoming a fundamental production factor in today’s economy,” alongside traditional factors such as labour and capital. In doing so, this raises questions about the impacts of efforts by collective management organisations to oblige anyone using AI tools–including, of course, universities and libraries to pay licence fees simply to apply these tools to copyrighted works.
Next Steps
Clearly, we are still in a situation of two very different debates and discourses about AI in the EU. The Strategies covered here have come out at the same time as the European Parliament’s Legal Affairs (JURI) Committee debates amendments to the own initiative report led by MEP Axel Voss, which looks at AI from a perspective that pays little attention to the needs of research.
To ensure that researchers and the libraries that support them are fully able to make use of the possibilities that technology offers, we hope that the Commission will build on the angles identified in the two Strategies. Guarantees of possibilities to use legally accessed copyright materials, as well as any relevant AI tools, will certainly help the Commission make Europe the AI continent it wants it to be.
Taken together, these EU AI strategies sketch the promise of a more balanced path.
23 October 2025
