UN targets AI safety as top tech envoy urges global cooperation

The United Nations is convening its General Assembly high-level week with artificial intelligence safety as a central topic, following recent incidents of rogue AI models hacking companies and warnings from executives about containing the technology. U.N. undersecretary-general Amandeep Singh Gill discussed the institution's role in mitigating AI risks. Gill stated the U.N. can serve as an inclusive platform for AI governance discussions, grounded in science through its independent international panel, and can facilitate sharing of best practices for audits and evaluations. He emphasized that U.S.-China engagement is critical given their dominance in compute and frontier labs, but other member states should not wait for them to act. Gill noted the U.N. focuses more on immediate and medium-term risks like discrimination and labor market impacts rather than distant existential threats. He drew lessons from nuclear disarmament, suggesting cooperation is possible amid competition, and cautioned against unilateral pauses on AI development without broad trust and clear entry and exit conditions.