The UN called for an international effort to prevent Artificial Intelligence from becoming uncontrollable

Likewise, he warned that A small group of people concentrates “almost unlimited” power on the development of AI and warned that the delay in establishing regulation mainly benefits large technology companies, their owners and those who support them.

In his speech during the inauguration of the 63rd session of the Human Rights CouncilAccording to the EFE agency, Türk said he shares the concern of industry experts about “the possible existential risk” posed by advanced AI.

He gave as examples cases of systems capable of escaping their test environments or of blackmailing their own developers to prevent them from being shut down, situations that, in his opinion, would demonstrate that the technology acquired excessive power.

This warning comes after several companies have detected, during security tests, behaviors of advanced AI models that They managed to bypass the limits of their security environments or that, in simulated scenarios, They resorted to deception and blackmail to avoid being replaced or turned off.

Türk acknowledged some recent progress in governance – the creation of the Global Dialogue on AI, the Independent International Scientific Panel on AI, the European Commission’s measures on platform liability – but considered that it is necessary to act more quickly.

To this end, the UN human rights chief asked countries that host AI development centers and those participating in their supply chains to agree clear and common boundariesdemand independent verifications and much greater industry cooperation on security.

Türk also noted that AI governance must address its effects on human rights, employment, democracy and the environment.

In this sense, he announced that he will soon write to companies in the sector to ask them to adopt measures under their control that will reduce the risks.