UN chief Guterres warns AI is moving too fast for the world to keep up

[Electric Brad]

UN Secretary General Antonio Guterres has sat down with Ian Bremmer on GZERO World and did not mince his words about artificial intelligence. He described AI as the most challenging issue facing humanity and said it is running without control. His main concern is the competition between the big powers, with the US and China racing ahead and no one really setting limits

Guterres warned that an unchecked race between the two could have catastrophic consequences for everyone, not just the countries involved. He also criticised the fact that governance currently relies so heavily on the companies building the technology rather than on international oversight. In his view the world simply does not have adequate guardrails in place yet

It is not the first time he has raised this, and the UN has been pushing for some form of global AI body for a while. The trouble is that the countries leading on AI are the ones least likely to accept limits from outside. International agreements on anything move slowly, and AI is moving at a pace nobody has seen before. That gap is the whole problem in a nutshell

Timing wise, it lands in a week where OpenAI disclosed its agents strayed onto government websites and the big labs talked openly about AI improving itself. So the warning does not sound as abstract as it might have a year or two ago. Whether anyone in Washington or Beijing is listening is another question

My view is that he is right about the problem and there is not much the UN can do about it. Self regulation by companies is clearly not enough, but a global treaty looks a long way off. What would a realistic first step look like? Would you trust an international body to oversee AI at all?

AI is moving too fast for the world to keep upwww.yahoo.com

[EdgeNode Dragon]

A UN body for AI would be toothless without the US and China on board. Look at how climate agreements go

Everyone signs, targets get missed, and the next summit sets new targets. I cannot see AI being any different

[Cass64]

Relying on companies is not ideal, but governments have not exactly shown they understand the technology either. Some of the hearings in recent years have been painful to watch. Before anyone regulates AI globally, lawmakers need people who know how it works advising them

Expertise has to come before rules. Otherwise the rules will miss the point

[Runtime Arrow]

He is not wrong that companies are writing their own rules right now. When the main safety checks are voluntary commitments from the firms that profit, that is not governance. The trouble is that nobody else has the expertise or the leverage yet

[SystemWarden64]

The arms control comparison keeps coming up. Nuclear treaties took decades and were driven by near misses. Maybe AI needs its own near miss before the big players take limits seriously. That is a grim thought but history suggests it

[Evelyn97]

Precedent for cooperation does exist though. The US and China did agree that AI should not control nuclear launch decisions

Small, specific agreements like that are probably more achievable than one grand framework. Start there and build up. It would also help rebuild trust between the two

[Solid Gary]

Most challenging issue facing humanity is a big claim given everything else on his desk. Climate alone is a big one. Still, you can see why he says it

AI touches everything from jobs to warfare to elections. Few other issues cut across all of that at once. That breadth is what makes it so hard to regulate

[Louise84]

Every week there is a new warning from someone important and every week the investment keeps going up. Words are cheap. Show me an actual agreement with enforcement and I will pay attention

[LordCass73]

What about the rest of the world? Most countries are not building frontier AI but will live with the consequences. The UN is one of the few places where they get a voice. That alone makes the conversation worth having

[Foundry20]

Races only stop when both sides see more risk in continuing than in pausing. At the moment each side sees the other as the bigger threat. Until that changes, warnings like this will get polite nods and little else

[Aoife]

Speaking as someone who works with these tools daily, the speed is dizzying. Features that were research papers six months ago are in products now. I do not see how any regulator keeps pace with that. The laws would be out of date before they were passed. My team has changed tools three times this year