Free beta access. 50 free AI tokens each week. Optional top-ups available. See what is included
Opening Word document…

NATO AI testing and standards solves integration with the EU

EU Institute for Security Studies 19, The European Union Institute for Security Studies (EUISS) is the Union’s agency dealing with the analysis of foreign, security and defence policy issues. Its core mission is to assist the EU and its member states in the implementation of the Common Foreign and Security Policy (CFSP), including the Common Security and Defence Policy (CSDP) as well as other external action of the Union. "The EU Nato and Artificial Intelligence" November 14. https://www.iss.europa.eu/sites/default/files/EUISSFiles/EU%20NATO%20AI%20-%20Report.pdf //pipk

On this basis, a number of specific areas for increased EU-NATO cooperation were discussed during the discussion. It was proposed that further steps were taken to implement the Joint Declarations, including possibly by establishing an ‘AI Centre of Excellence’. Another proposal concerned adding an AI research dimension to the Declarations, although a different view contended that AI would nonetheless automatically and incrementally feature in EU-NATO cooperation given its crosscutting nature. The need for a common EU-NATO data-management and data-sharing framework was also proposed, in light of the importance of data pools for the performance of AI algorithms. Although much remained to be done in that area, data labelling was identified as a good starting point.

Another point raised on several occasions was the importance of interoperability. Here, it was argued, AI had the potential to either increase or decrease existing gaps. Militaries could mitigate AI interoperability gaps through exercises and common training, such as through the Cyber Education Platform, or by developing niche capabilities, especially by utilising EU initiatives such as Permanent Structured Cooperation (PESCO) and the European Defence Fund (EDF). Some participants argued that divergences were nevertheless inevitable and that, consequently, countries with low or no AI capabilities should be able to trust the technologies of those with more advanced technologies. This, in turn, required a common framework for certification and evaluation procedures, as well as standardised interfaces and modular system architectures that would combine closed/‘national’ and open/‘shared’ components.

Sign in to recut, update, or save this card.