In September 2025, the UN General Assembly held its 80th session in New York. Among the many tensions present at this meeting, a call was made regarding the use of AI and its regulation.
A group of NGOs and scientific experts have sounded the alarm with a clear message: it is imperative to quickly establish a global legal framework to regulate artificial intelligence.
This technology is evolving without any boundaries, with each country acting according to its own rules. The result ? A disorderly mosaic of regulations in the face of AI, which knows no geographical limits.
Heavyweights of science behind the appeal
We’re not talking about just anyone. Among the signatories are Yoshua Bengio and Geoffrey Hinton, two legends in the field of AI, as well as Nobel Prize winners in economics Joseph Stiglitz and Daron Acemoglu. It would be hard to find a more credible group.
The initiative comes from three specialized organizations: the Center for the Safety of AI (CeSIA), The Future Society, and the Center for Human-Compatible AI at Berkeley. These organizations represent the “Doomers” camp, those who are seriously concerned about the existential risks of AI. They are the opposite of the “accelerationists” who want to rush headlong into technological development.
Target: red lines before the end of 2026
The goal is to reach an international political agreement that sets clear and unbreakable limits before the end of 2026. With AI permeating every aspect of our lives, from healthcare and education to finance and transportation, prevention is better than cure.
In short, this means leveraging the advantages of AI while controlling its risks to avoid a dangerous backlash.
The initiatives already in place to regulate AI
A few initiatives already exist around the world, forming a regulatory mosaic around AI.
Europe has been a pioneer with the adoption of its AI Act in June 2024, establishing a clear regulatory framework for artificial intelligence. This regulation prohibits practices of concern such as Chinese-style social scoring, widespread mass surveillance, and manipulative AI systems, while strictly regulating uses deemed “high risk.”
In 2019, the OECD established guiding principles on AI, which were recently updated to better address the specific challenges of generative AI. At the same time, the Global Partnership on Artificial Intelligence (GPIA) brings together international actors. These initiatives provide a solid foundation for a future harmonized international framework.
However, the main challenge remains the current fragmentation: each country still adopts its own regulatory approach, while AI circulates freely across borders.
When geopolitics complicates everything
On paper, global governance of AI seems logical. In real life, it’s a different story.
Not all countries are on the same page. Some see AI as a threat that needs to be controlled, while others see it as an economic gold mine that should not be restricted.
The head of United States, Donald Trump has even threatened sanctions against countries that regulate AI too heavily. This is understandable, as almost all of the AI giants are American. For Washington, it’s a question of technological dominance and sovereignty.
A race against time
The appeal launched at the UN is much more than a publicity stunt. It is a real race against time: AI is advancing at breakneck speed while the United Nations drags its feet in reaching an agreement.
The real question is whether the United Nations will be able to agree on common limits before AI becomes uncontrollable. Unfortunately, the experience of international negotiations on sensitive technologies does not inspire optimism.
However, as this appeal points out, some of the risks associated with AI are too great to be ignored.
References :
- https://www.lemonde.fr/pixels/article/2025/09/22/intelligence-artificielle-un-appel-mondial-pour-fixer-des-lignes-rouges
- https://www.un.org/fr/global-issues/artificial-intelligence
- https://fr.wikipedia.org/wiki/PMIA
This article is really interesting; it’s true that the speed at which AI is evolving can be frightening. But even with the best intentions, rogue states and criminal groups will misuse it.