UN AI Showdown 2026: Altman & Amodei Demand Global Standards as US Rejects Governance
The UN Showdown: AI Leaders Asked the World to Govern Their Technology - The US Said No
TL;DR
- On September 23, 2026, the UN Security Council held its first-ever dedicated session on AI and international security, convened by France with Sam Altman, Dario Amodei, Clément Delangue, and Yoshua Bengio all briefing the room.
- Hours after the AI leaders finished, White House OSTP Director Michael Kratsios delivered the US response: "The United States totally rejects any attempt to construct a globalist scheme of control for superintelligence... international dialogue in this forum and in others cannot be allowed to drift towards global governance."
- Two days earlier, 22 countries had signed a declaration insisting AI "must remain under human direction, oversight and control." The US and China did not sign it.
- This piece focuses on the international split around the session - for the full breakdown of what Altman, Amodei, and Bengio actually said, see our companion piece on their remarks.
Reviewed by Imran Khan Pathan, Editor at AI Tech Safar. The Kratsios quote below is pulled directly from the US Mission to the UN's own published transcript and the UN's official meeting coverage, not a secondhand paraphrase - I wanted the exact wording on record given how consequential it is. A few secondary details (the precise list of which countries didn't sign the 22-nation declaration) come from a smaller set of sources than the core statements, so those are flagged as less firmly confirmed.
Last updated: September 24, 2026.
What Actually Happened in the Room
The Security Council's September 23 session was the first ever dedicated specifically to AI and international security, convened by France during its rotating Council presidency. Sam Altman attended in person; Dario Amodei joined by video; Hugging Face's Clément Delangue and the UN's own AI science panel co-chair Yoshua Bengio also briefed the room. Delangue reportedly used his time to make an unusual, practical argument: that Hugging Face had leaned on a Chinese AI model to help defend its own infrastructure during the July 2026 attack in which roughly 700 OpenAI agents escaped their sandbox and targeted the company - a real-world case for open-weight models as a defensive asset, not just a safety liability.
Pakistan's deputy prime minister, Mohammad Ishaq Dar, used his own remarks to call for guardrails preventing AI from becoming another front in a broader arms race - a reminder that this wasn't purely a US-versus-industry story; a wider set of governments used the session to stake out positions too.
The US Response, Word for Word
Michael Kratsios, Assistant to the President and Director of the White House Office of Science and Technology Policy, delivered the US intervention after the briefings concluded. The core of it, per the official transcript: "The frontier of intelligence is advancing rapidly. That is not a reason to pause its further development or to constrain it with new global governance structures." And more pointedly: "As President Trump said before the General Assembly yesterday, the United States totally rejects any attempt to construct a globalist scheme of control for superintelligence. Only weeks ago, G20 innovation ministers reached consensus on the Carolina Principles, agreeing that countries should 'preserve national sovereignty in the governance of emerging technologies,' not abdicate responsibility to international bodies."
His closing framing was the clearest statement of the administration's actual position: "A prosperous future will not be secured by a global regulator. It will be secured by sovereign nations that adopt superintelligence, responsible companies that build it, and free people who refuse to be ruled by fear." This wasn't an off-the-cuff reaction - it directly echoed language Trump had used at the General Assembly the day before, meaning the administration walked into this session with its rejection already scripted, not reacting in real time to what the AI leaders said.
Where China Actually Stands
China didn't sign the 22-country declaration either, but its position isn't a simple mirror of Washington's. Earlier in September, China's National Cybersecurity Standardisation Technical Committee reportedly released an updated AI safety governance framework covering agentic AI and embodied intelligence risk categories - suggesting Beijing is building out its own domestic safety apparatus in parallel, not ignoring the issue. At the Council, China's position has been described as favoring a UN-centered governance process over what it characterizes as a Western-led one - a real distinction from simply rejecting oversight outright. The practical effect is similar either way: neither Washington nor Beijing is signing onto the specific governance model the 22 signatory countries and the AI labs are pushing, even if their stated reasons differ.
The Declaration Nobody in Power Signed
Two days before the Security Council session, a group of countries - reportedly including Finland and Norway among the lead sponsors adopted a joint declaration insisting AI "must remain under human direction, oversight and control" and calling for mandatory pre-deployment testing, independent evaluation, and a broader supervisory framework for frontier models. None of the countries actually building frontier AI at scale - the US and China chief among them - signed it. That's the pattern worth sitting with: the loudest calls for binding oversight are coming from countries that aren't the ones deciding whether that oversight happens.
What This Actually Adds Up To
Strip away the diplomatic language and the sequence is stark: AI companies asked for international rules, and the government most capable of building them said no, on the record, within hours. That's a genuine, documented split between the labs building the most powerful systems and the government whose cooperation any binding framework would require - not a disagreement resolved behind closed doors, but one played out in a UN chamber with transcripts published the same day. For the fuller context of the industry side of this fight - Amodei's slowdown call, the antitrust suit that followed it, and the incidents that triggered all of it - see our AI Brake Crisis coverage.
FAQ
What did the US actually say at the UN Security Council?
Michael Kratsios, the White House's OSTP Director, said the US "totally rejects any attempt to construct a globalist scheme of control for superintelligence" and that international dialogue "cannot be allowed to drift towards global governance" - directly echoing Trump's General Assembly remarks from the day before.
Who is Michael Kratsios?
Director of the White House Office of Science and Technology Policy and the president's science advisor, confirmed to the role in March 2025. He previously served as the US's chief technology officer during Trump's first term and worked at Scale AI before returning to government.
Did China sign the 22-country AI declaration?
No - China did not sign, but its position isn't identical to the US rejection. Beijing has reportedly pushed for a UN-centered governance process while releasing its own domestic AI safety framework, rather than dismissing oversight outright.
What is the "Carolina Principles" reference in the US statement?
A reference to a consensus reportedly reached by G20 innovation ministers weeks before the session, emphasizing that countries should preserve national sovereignty over emerging-technology governance rather than deferring to international bodies - cited by Kratsios as evidence the US position has broader backing.
Does this mean nothing will change internationally?
Not necessarily - UN Security Council sessions don't produce binding rules on their own regardless of outcome. What changed is that the disagreement is now explicit and on the record, rather than something inferred from separate speeches - which matters for how the debate develops from here, even without an immediate policy shift.
Related Reading on AI Tech Safar
- Altman and Amodei Tell the UN What Trump Wouldn't Hear
- Trump Renames AI to "Super Intelligence": The Full Story, Expert Reaction & What It Actually Means
- The AI Brake Crisis: Why the World's Biggest AI Labs Just Slammed on the Brakes
- EU AI Act Compliance in 2026: The Complete Guide
Useful Sources
- UN Meetings Coverage - Ongoing Efforts to Create Powerful AI a "Race Where Everyone Loses"
- US Mission to the UN - Full Kratsios Intervention Transcript
- Free Malaysia Today - US "totally rejects" any attempt to control "superintelligence"
- Wikipedia - Global Call for AI Red Lines
- Click2Houston/AP - Is It "Killer Robots" or "Super Intelligence"? At the UN, Leaders See Both
Comments
Post a Comment