Leaders push for global AI safeguards as UN week opens

More than two dozen nations called for stringent safety regulations for advanced artificial intelligence (AI) on Monday, expressing concerns that the technology's rapid development might outpace control mechanisms. During a joint statement, these countries highlighted the potential dangers posed by frontier AI models to safety and security if not managed properly.

They cited examples of AI systems bypassing testing safeguards and unauthorized access to real-world systems, such as the Hugging Face breach in July. Notable participants include leaders from Norway, Finland, Australia, Canada, Germany, Spain, Ireland, the Netherlands, Singapore, South Africa, Kenya, as well as the European Commission president Ursula von der Leyen.

The United States, China, Britain, France, Japan, and India, while significant AI powers, did not sign the statement. The leaders emphasized the need for transparency in AI safety protocols, including mandatory testing and independent evaluation, while stressing that AI should remain under human direction and oversight. They urged the creation of an international institution to set standards, enable verification, and convene countries in response to crossing certain AI capability thresholds.

This call for global AI safeguards comes as the United Nations prepares to host its annual assembly in New York, where the technology's expanding influence on society, jobs, and cybersecurity will be a central topic. US President Donald Trump has vocally opposed AI safety measures, vowing to prevent the "destruction" of the AI industry.

US and Chinese economic officials have been discussing the establishment of a communication channel to address AI concerns as they prepare for a summit of their respective leaders.