The Independent International Scientific Panel on Artificial Intelligence within the United Nations said that safety measures should not be postponed until the risks of advanced artificial intelligence agents are understood. Established last year, the panel’s thematic assessment examines control and safety issues following the Hugging Face attack linked to OpenAI that came to light earlier this year. According to the report, governments should take early action rather than wait for incidents; scientific uncertainty should not delay regulation. As the capabilities of autonomous and semi-autonomous agents expand, international coordination and shared principles of safety and responsibility should be strengthened; resources should be allocated to security and accountability mechanisms. In the assessment published during the UN General Assembly session in New York, António Guterres called on countries to act together. Guterres said that a race to the bottom on safety standards was unacceptable. The panel considers loss of control an uncertain but high-consequence risk.
Why it matters
The central issue in the assessment is whether safety decisions can be postponed until conclusive scientific evidence is available as the capabilities of artificial intelligence agents expand. This framework positions artificial intelligence governance not merely as a matter of technological progress, but as an issue of early action, shared standards and shared responsibility. The recommendations place at the center of international coordination the question of how differing safety approaches among countries will be harmonized in terms of control, resource allocation and accountability. However, the uncertainty surrounding the risk of loss of control leaves open questions about what measures should be made mandatory at which threshold and how responsibility for serious consequences should be distributed.
Term: agent
An artificial intelligence agent is software that uses tools and carries out multi-step tasks to achieve a goal rather than producing a single response.