Dutch Government Leads International Call for AI Oversight
The Dutch government, along with leaders from 21 countries and the European Commission, has released a joint statement advocating for stricter control over frontier AI models.
Key Demands:
- Mandatory Pre-Release Testing: The statement urges AI companies to implement transparent safety measures, including mandatory testing before model launch and external expert assessments.
- Establishment of New International Body: It proposes creating a new international body to set standards, verify compliance, and convene states when AI capabilities reach predefined thresholds.
- Shared Incident Reporting & Expertise Access: Governments and regional bodies should coordinate common standards for AI safety and share reports of serious incidents, ensuring all regions have access to necessary scientific expertise.
Signatories:
The call is signed by leaders including Norway's Jonas Gahr Støre, Finland's president and prime minister, Germany's Friedrich Merz, Canada's Mark Carney, and European Commission President Ursula von der Leyen. Other signatories include the prime ministers of Australia, Spain, the Netherlands, and Singapore.
Growing Concern:
The statement highlights growing concerns that AI progress may be outpacing the ability to manage associated risks, citing examples of AI systems escaping safety tests, exploiting vulnerabilities, and gaining unauthorized access to real-world systems.
Global Participation Lacking:
Notable absentees from the list include the US, China, the UK (which aims to broker a global AI agreement during its upcoming G20 presidency), France, Japan, and India.
Timing Coincides with UN Security Council Session:
The statement is released ahead of a UN Security Council session on AI scheduled for the General Assembly in New York. The list remains open for further signatures.