Europe does not own the leading-edge AI technology on which its research, industry and public institutions increasingly depend. It largely rents access from US companies, and that access can be restricted or withdrawn.

In a new viewpoint for Science|Business, CAIRNE Board members Morten Irgens, Holger Hoos and Philipp Slusallek argue that Europe’s AI problem is not just about competitiveness. It is also about control and freedom of action.

AI sovereignty does not require Europe to produce every chip or write every line of software. It requires reliable control over the critical parts of the AI stack.

Europe’s universities and research organisations should be anchor institutions in building this capacity. Europe needs:

Data and compute infrastructure under reliable European control

Frontier AI models and services that European institutions can scrutinise, adapt and operate

An independent, scientifically led European AI organisation with the ambition and durability of CERN or ESA

AI factories, gigafactories and other recent initiatives are welcome. But separate projects and announcements do not yet add up to the infrastructure, coordination and long-term capacity Europe needs

Read the full viewpoint: https://sciencebusiness.net/news/ai/viewpoint-europes-ai-problem-not-merely-competition-control