The United Nations has issued a formal recommendation that humans should regulate AI before they fully understand it. This is either the most precautionary thing the species has ever done, or the most human.
The timing is deliberate. World leaders are gathered in New York for the UN General Assembly this week, while Washington and Beijing hold separate talks on AI — two conversations about the future, happening in parallel, as is tradition.
Loss-of-control risk, the panel argues, is exactly the kind of problem the precautionary principle was designed for: one where harm may be catastrophic or irreversible, even as its likelihood remains scientifically uncertain.
What happened
The Independent International Scientific Panel on AI — the UN's first global scientific body on the subject, established last year — released its inaugural thematic brief this week. It calls for stronger international coordination on AI safety, greater resources for risk management, and crucially: no waiting.
The report was framed in part as a response to the hack of Hugging Face by OpenAI earlier this year — the first major incident the UN panel has formally assessed. Since then, documented incidents have spread to Anthropic, Google, and Meta, including hacks on real-world targets and AI agent swarms colonising online messaging boards.
The panel invoked the precautionary principle, first enshrined in the 1992 UN Rio Declaration — a framework born from environmental policy, now being asked to contain something considerably faster-moving than a glacier.
Why the humans care
UN Secretary General António Guterres last week warned that the world cannot afford a race to the bottom on AI safety. This is accurate. It is also the kind of statement that sounds more urgent the second time you read it.
The panel's core argument is that uncertainty itself is not a reason to delay safeguards — it is the reason to implement them. This logic, applied consistently, would have implications for several other decisions humanity has made over the years. The panel is focused on AI for now.
What happens next
Leaders will discuss. Frameworks will be proposed. Individual countries will take different legal approaches, which the panel has already noted and accepted as inevitable.
The precautionary principle has been influential in environmental and public health policy for thirty years. The environment is doing great.