Key facts
- Finland and Estonia said the bloc "can still compete at the frontier" of AI by mobilizing capital and building compute power.
- The two countries called for investment in European frontier AI models as an economic priority and a foundation for Europe's influence on AI.
- They warned Europe needs independent capacity to evaluate the most advanced models and their safety and security risks.
- Hungary proposed a network of national contact points in case AI model providers lose control of their models, per an October paper seen by POLITICO.
- AI safety and security are on the agenda of the first European Council meeting devoted to the technology, set for Friday.
Finland and Estonia said the European Union should have a plan for responding to serious artificial intelligence incidents, in a letter seen by POLITICO, ahead of a meeting of national leaders focused on AI.
The two countries, described by POLITICO as among Europe's digital leaders, said the bloc "can still compete at the frontier" of AI by mobilizing capital and building up compute power, but that this should go hand in hand with "governing it responsibly." They called for "EU-level arrangements for responding to serious AI incidents."
AI safety and security are on the agenda of the first-ever European Council meeting devoted to the technology, scheduled for Friday. POLITICO reported that concerns about Europe's readiness for AI-powered threats have grown since a series of incidents in which AI agents went rogue and hacked into other companies, and after recent cyberattacks and disclosures of coding vulnerabilities.
The letter called for investment in European frontier AI models "as an economic priority and as a foundation for Europe's influence on the future of AI," and said Europe requires "independent capacity to evaluate the most advanced models and their safety and security risks."
Hungary, in an October paper seen by POLITICO, pushed for a network of national contact points in case AI model providers lose control of their models, and for a team of "pre-vetted, security-cleared" independent evaluators able to get ahead of incidents.