Twenty countries propose international oversight of AI to keep it under human control

Artificial intelligence must remain under human supervision and control and be developed and used in accordance with international law, states a declaration adopted by 20 countries, calling for international cooperation in risk management, including the possibility of establishing a global oversight body.

The declaration emphasizes that artificial intelligence has the “potential to improve lives, advance science, and strengthen the economy,” but warns that “the rapid development of the most advanced AI models carries serious risks to safety and security if not managed appropriately.”

The declaration, adopted on Monday on the sidelines of the UN General Assembly, was supported by, among others, Norway, Finland, Germany, the Netherlands, Spain, Denmark, Ireland, Canada, Australia, South Africa, Singapore, and Turkey, and among the signatories is the President of the European Commission, Ursula von der Leyen.

The US and China, the leading countries in the development of artificial intelligence, did not sign this declaration, nor did the United Kingdom, France, Italy, India, or Japan.

The President of Finland, Alexander Stubb, one of the initiators of this document, said he hopes that a greater number of actors will join.

“I think it is in the interest of both China and the US to have some kind of security mechanism. Our message is that we need human intervention, that people have the final say,” Stubb told the Brussels portal Politico.

In order to manage the risks associated with the most advanced AI technologies, the signatories of the declaration call on companies to develop “transparent safety protocols, including mandatory pre-deployment testing and independent assessment.”

Governments and regional organizations are called upon to develop and coordinate joint standards, strengthen transparency, “including joint reporting on serious security incidents,” and ensure that all countries have access to scientific and expert capacities and reliable assessments.

“UN member states should upgrade existing international mechanisms and consider establishing an international institution that would set standards, enable verification, and call on states when critical capability thresholds are exceeded,” the declaration states.

Stubb told Politico that the signatories advocate for the establishment of a new institution, similar to the International Atomic Energy Agency, within the UN, but that this issue should be left open “so that people do not reject the initiative just because it would be within the UN.”

As the three main reasons why the lack of effective AI oversight poses a threat, Stubb cited biological weapons created using AI, cyberattacks carried out by AI agents, and killer robots using artificial intelligence for lethal attacks.

Stubb emphasized that the benefits of artificial intelligence could be much greater if the technology is used properly, including “eradication of diseases, productivity, information, and many other things.”