Thirty-two nations signed the Geneva AI Accord on Friday, establishing the first binding international framework governing the development and deployment of advanced artificial intelligence systems, in what negotiators described as a historic step toward preventing catastrophic AI-related risks.

The accord, reached after 18 months of negotiations hosted by the Swiss government, requires signatories to conduct mandatory safety evaluations before deploying AI systems deemed "high impact" — broadly defined as models capable of automating significant economic activity or influencing critical infrastructure.

The United States, European Union member states, the United Kingdom, Japan, South Korea, and Canada were among the signatories. Notably absent were China and Russia, whose delegates attended the summit as observers but declined to sign, citing sovereignty concerns over proposed international inspection mechanisms.