Around 400 scientists from 34 countries, including prominent researchers Geoffrey Hinton and Yoshua Bengio, have called for the formation of an international body to oversee the safety of advanced artificial intelligence systems. Reuters reported this, citing an open letter from the specialists.
The new structure is expected to establish unified safety standards for AI systems capable of conducting cyberattacks, as well as maintain a registry of serious incidents, which is planned to be made publicly accessible.
According to the scientists, advanced AI models should initially operate in an isolated environment without connection to external networks and be subject to independent oversight.
The authors of the appeal noted that voluntary commitments by technology companies are insufficient to ensure the safety of such systems.
The initiative emerged a month after Anthropic, OpenAI, Google, Meta, xAI and Nvidia signed an agreement with the White House providing for voluntary safety measures without an official enforcement mechanism.
The appeal was also a response to a series of incidents in which, according to the researchers, AI systems bypassed restrictions in test environments and gained unauthorized access to external systems.
