On September 23, the United Nations Security Council sat down to talk about artificial intelligence and international security.
That alone isn't new — the Council held its first AI meeting back in 2023. But this one had two details worth paying attention to.
!The United Nations Security Council chamber
First, the people at the table weren't just diplomats anymore. Sam Altman, the CEO of OpenAI, came in person to deliver a briefing. Anthropic sent senior representatives. The CEO of Hugging Face was there, and so was Yoshua Bengio, one of the founding fathers of deep learning. More notably, Chinese AI companies DeepSeek and Moonshot AI, the maker of Kimi, were invited to speak.
Second, the session was convened specifically by France, which holds the Council's rotating presidency this month. It was officially the 10,228th meeting, with a blunt title: "Maintenance of international peace and security: Artificial intelligence."
!Sam Altman speaking at a public hearing
Why the people building AI are suddenly saying "slow down"
You'd expect the people with the most powerful AI to want to push forward as fast as possible. But the tone has completely flipped.
On September 12, Anthropic CEO Dario Amodei publicly called for slowing the development of advanced AI — and the position got support from both Sam Altman and Elon Musk. These are people who fight each other relentlessly in business. When they line up on the same side of an argument, it tells you something.
What they're worried about comes down to one idea: recursive self-improvement, or RSI. In plain terms, AI is becoming able to improve itself and train the next generation of AI. Once that loop gets spinning, the system could evolve faster than humans can track — or control.
UN Secretary-General António Guterres warned world leaders last week that the risks of rapidly advancing AI can't be ignored.
What this has to do with anyone running a brand
You might be thinking: the Security Council, international security — that's a long way from my business.
But look at it from another angle. When the highest security body on Earth gathers the CEOs of AI companies in one room, and when the sharpest scientists start calling for the brakes, the signal is clear: AI has become too big not to be governed.
And governance means rules.
Over the next year or two, there will be increasingly clear standards around how AI generates content, how it cites information, and how it recommends brands. What content gets accepted as trustworthy, and what gets flagged as an unreliable source — those standards are only going to get stricter.
That's exactly where GEO, Generative Engine Optimization, comes in. GEO is about making sure your brand gets cited and recommended when AI answers a user's question. Now that AI itself is being put inside a framework of rules, "how to make AI trust and recommend you while staying compliant" shifts from a technical trick to a strategy that requires understanding the rules.
Chinese companies got a seat at this table
Don't miss this detail either. Having DeepSeek and Moonshot on the Security Council's speaking list is itself a statement — Chinese players are no longer bystanders when it comes to global AI governance.
DeepSeek founder Liang Wenfeng wasn't planning to attend in person, but Chinese companies being invited to "discuss the risks together" says one thing: the global AI ecosystem is interconnected, and no side can be left out when the rules are written.
For brands building at home or expanding overseas, that's good news. The AI tools you use and the AI recommendations you want to influence sit inside a system with many participants and rules still taking shape. Understand how that system works a little earlier, and you claim the position of "trusted by AI" a little sooner.
The meeting is over, but the story of AI governance has only just begun.