EU agencies begin testing Chinese open-source AI models

The European Union Agency for Cybersecurity (ENISA) and the Joint Research Centre (JRC) have begun testing open-weight AI models from China. This initiative is part of a broader comparative study that includes evaluating leading models from the United States and Europe to analyze their capabilities, performance, limitations, and potential scientific applications.

European Commission spokesperson Maciej Berestecki clarified that the JRC is not negotiating for exclusive access or licenses with Chinese developers. Instead, the models are being run on secure, internally hosted servers. The testing does not focus on AI safety or cybersecurity protections, but rather on scientific assessment.

The move highlights Europe's current struggle to develop its own cutting-edge AI models. While European companies like Mistral are launching advanced systems such as Mistral Large 4 to compete, some European firms are reportedly considering Chinese open-source alternatives. This dynamic places European public and private organizations in a position where they may have to choose between American and Chinese technology providers for advanced applications.