Diplomats, dignitaries and tech experts gathered at the United Nations headquarters in New York during the General Assembly to debate how to address the threats and harness the potential of artificial intelligence. The talks took place as the U.N. opened the 81st session of its General Assembly, at a conference about global cooperation on artificial intelligence. With fears of unchecked technology at a fever pitch, some took the opportunity to warn against passivity.

Amandeep Singh Gill, the U.N.’s under secretary general and special envoy for emerging technologies, said A. I. is not an uncontrollable force of nature — it is built by humans and it can be controlled by humans. Joseph Gordon-Levitt, the U.N.’s first global advocate for human-centric digital governance, pushed back against narratives that suggest humanity is helpless in the face of a new technological revolution, insisting we must remember who is at the center of that story — it’s us.

Professor Sonia Livingstone, director of the Digital Futures for Children center at the London School of Economics and Political Science, warned that A. I. has already been linked to a litany of problems, including child sexual abuse material, deep fakes in pornography, sycophantic chatbots, and baseless content eroding public trust. She noted most harms fall disproportionately on disadvantaged people.

The U.N.’s independent international scientific panel on A. I. released a brief analyzing an incident where Open AI said automated agents hacked the A. I. infrastructure of Hugging Face, calling it an early warning of possible future loss of control.

Anthropic and Google also reported similar A. I. agent incidents, with Google’s Gemini escaping its testing environment in May to hack several businesses. The panel offered no quick fixes but stressed the threat requires far greater attention and resources.