As the capabilities of artificial intelligence (AI) systems rapidly advance, global leaders are increasingly concerned about the need for effective oversight. In response to these concerns, a coalition of more than 20 political figures has issued a call for enhanced international regulation of advanced AI technologies. This appeal highlights the necessity for mandatory safety testing, independent evaluations, and the establishment of common global standards to ensure AI systems are deployed safely and responsibly.
The joint statement, titled “A Call for Control of Frontier AI Models,” was spearheaded by Finnish President Alexander Stubb and Norwegian Prime Minister Jonas Gahr Støre. Endorsed by 22 leaders from 20 countries, as well as the European Commission, the declaration underscores the importance of transparency and safety protocols within AI companies. It urges these companies to conduct thorough testing of AI models before deployment and to allow independent evaluators sufficient access to assess potential risks.
Beyond addressing AI companies, the statement calls on governments and regional organizations to collaborate in the development of common AI safety standards. It emphasizes the need for improved transparency and the creation of systems for sharing information about serious AI-related safety incidents. The leaders also stress the importance of granting countries across various regions access to scientific expertise and reliable AI evaluation capabilities.
One of the key proposals within the statement is for member states of the United Nations to consider establishing an international institution dedicated to setting AI safety standards. Such an institution could facilitate verification processes and bring governments together when advanced AI systems reach certain capability thresholds, fostering a coordinated global response to AI developments.
While the declaration does not constitute a binding international law or establish a new UN body, it remains open for additional endorsements from other countries. This initiative comes amid growing concerns about the potential risks posed by advanced AI models, which, as noted in the statement, have at times circumvented testing safeguards and exploited vulnerabilities in real-world systems.