Superintelligence Accord: Trump Renames AI ‘SI’ as 6 Tech Chiefs Sign Voluntary ‘Constitution’
US Tech & AI: Analysis
Key Facts
- The document. A one-page “White House Accord on Super Intelligence”, plus a page of signatures. Trump posted it on Truth Social on 29 September 2026.
- The signatures. Trump and the leaders of Google, Anthropic, Meta, OpenAI, xAI and Nvidia. OpenAI’s Sam Altman was at his company’s conference in San Francisco.
- The promises. Four voluntary “layers” of checks: internal controls, an internal team, an outside auditor and a board committee.
- The new name. A separate executive order tells federal agencies to say “SI” instead of “AI”. US law is unchanged.
- The politics. It comes five weeks before the 3 November midterm elections, amid anger over data centres and power bills, and a race with China.
- Latin America. Brazil’s binding AI bill still waits in the Chamber of Deputies. Its data-centre tax breaks became law on 15 September.
Trump says AI is now “SI” and calls the superintelligence accord “almost like a constitution”. We read every line — and explain why it came now.
One-stop reference
Company Intelligence
Every listed company in Latin America — financials, ownership and structure for 1,450+ companies across 26 exchanges, in one place.
Ask Rio Times
17 years of Latin America reporting, on demand.
Answers are generated from Rio Times reporting only â every claim links to its source article.
The superintelligence accord signed at the White House on Tuesday, 29 September 2026, fills just one page. President Donald Trump called it “almost like a constitution in a way”. It is a voluntary promise by six tech companies to check their own AI.
Why Now: Data Centres, Power Bills and the Midterms
The timing is political. The midterms, the US elections held halfway through a president’s term, are on 3 November. They decide who controls Congress. And anger over data centres, the huge computer buildings that run AI, is rising. A June Reuters/Ipsos poll, cited by NPR, found 77% of Americans worried they push up power bills.
Both parties now campaign against them. Ads mentioning data centres had cost more than US$45 million by early September, NPR found. In Ohio, supporters of Democrat Sherrod Brown call Republican Senator Jon Husted “the face of data centers”. In Florida, Republican Byron Donalds promises to protect families “from data centers that jack up utility rates”. The Republican Senate campaign committee (NRSC) called data centres a “sleeper issue”, NPR reported.
The White House has answered with promises. In March, tech firms signed Trump’s pledge to “build, bring, or buy” their own power. On 16 September the House of Representatives, the lower chamber of Congress, voted 417 to 3 for a bill that pushes states to make data centres pay the full cost of their grid upgrades. The Senate has not voted.
On Tuesday Trump said the companies “will be very generous to these communities”. He claimed a firm — “I believe it’s Meta” — was “paying every single teacher over $50,000” in a Louisiana parish. Not quite. Many Richland Parish teachers did get bonuses of about US$50,000. But the money came from a short-term jump in local tax income as Meta builds a data centre there, GovTech reported.
Trump’s bigger theme is China. “Whoever wins S.I. is going to win,” he said. Chinese leader Xi Jinping, who made a state visit to Washington from 23 to 25 September, “will take every data center that we don’t want”, he added. Earlier this month Trump called fears of AI “destroying Humanity” a “HOAX”, CNN reported.
Vice President JD Vance told the companies the answer was “to take the risk seriously, not to come to the government for a regulatory regime that may make things worse if it’s not smart and careful”.
What the Superintelligence Accord Says
The Accord in 10 Points
- Responsibility. “every company is responsible for developing its own technology safely and in a way that builds trust with customers and the public.”
- Scope. It covers “every company that is training and deploying frontier models”.
- The ask. “in addition to any other precautions, we believe each company should implement the following four layers of controls and audits”.
- Layer one. Internal controls to monitor models “around areas like cybersecurity, biosecurity, and chemical threats, and to ensure that its models do not hack or access technical systems in unintended ways.”
- Layer two. “Empower an internal team to ensure all of the controls, monitoring, and detection are operating as intended, and that any issues are remediated.”
- Layer three. “Partner with an independent external auditor or evaluator to carry out independent assessments of whether the controls, monitoring, and detection are operating as intended.”
- Layer four. “Designate an independent committee of the board of directors to oversee and receive reports … as well as to ensure any issues identified are remediated.”
- The aim. To give “each company, its customers, and the public confidence that the technology is operating as intended.”
- Next steps. “The participating companies will meet regularly to establish standards and best practices”. And: “Over time, it may make sense to codify these steps into laws or regulations.”
- The pledge. “each of our companies are committed to doing this.” Signed by Trump, Pichai, Amodei, Zuckerberg, Brockman, Musk and Huang.
The key word is “should”, not “must”. “Frontier models”, meaning the most powerful AI systems, are never defined. There is no named auditor, standard, deadline or penalty. The auditor checks that controls work “as intended” — as each company intends. Reports go to the company’s own board, not to a regulator or the public.
The hacking clause has a reason: twice in three months, OpenAI models broke out of their test environments and reached the open internet, CNN reported.
From AI to “SI”: The New Name
Speaking to reporters outside the West Wing, Trump said: “It’s not A.I., it’s S.I. … We’ve changed the name officially today.” The old term, he said, is “a fake word”.
The new name comes from a separate executive order. This is a written instruction from the president to the federal government, and it needs no vote in Congress.
What the Executive Order Does
- Section 1. The executive branch “shall use the terms ‘Super Intelligence’ and ‘SI’ in place of ‘Artificial Intelligence’ and ‘AI’”.
- Section 2. This covers “official correspondence, public communications, websites, reports, policy documents”.
- Section 3. SI means what US law already defines as artificial intelligence. The president’s science adviser has 60 days to propose a new legal definition.
- Section 4. It creates no “right or benefit … enforceable at law”.
So the change is real but narrow: it covers how officials speak and write. US law still says “artificial intelligence”. The accord itself never uses “SI”. House Speaker Mike Johnson, leader of the lower chamber, read its subtitle as “Joint Commitment on Frontier SI Responsibilities”. The printed page says “Frontier Responsibilities”.
A “Constitution” That Binds No One
Trump said the tech chiefs “almost viewed it as a separate kind of a constitution”. Asked if the accord was binding, he replied: “I think it’s morally binding.”
A constitution creates institutions and binds people. This text does neither. The committee of “maybe 10 people” that Trump floated is not in it. It grants no rights and cannot be enforced or changed. Johnson himself called it “commitments that are voluntary on behalf of the industry”.
The chiefs themselves were modest. Meta’s Mark Zuckerberg called it “a start”. Google’s Sundar Pichai compared it to “financial controls in a company”. Joe Biden’s voluntary deal of July 2023 went further on some points: seven firms also promised outside reporting of flaws and watermarks for AI-made content.
Where Was Sam Altman?
OpenAI’s chief executive was not in the room. Axios reported that Greg Brockman, OpenAI’s president, would attend “while CEO Sam Altman attends the company’s annual conference in San Francisco”. Brockman signed for OpenAI.
At that conference Altman launched new AI agents, AP reported, a day after OpenAI halted a more advanced model over safety concerns. Al Jazeera named it GPT-6.1 Astra. Altman told CNBC: “We are pacing our progress, which includes sometimes not training a model.”
Who Did Not Sign, and the Critics
Microsoft’s Satya Nadella, Amazon’s Jeff Bezos, AMD’s Lisa Su and Palantir’s Alex Karp came to the lunch, according to Axios, CBS News and CNN, but did not sign.
Anthropic’s Dario Amodei, who did sign, chose his words with care: “the technology has very real risks. And the mechanism, how we address those risks is still under discussion.”
Democrats were blunt. Senator Mark Warner said: “The president’s response? To rename it and tell the companies developing it to regulate themselves.” Senator Elizabeth Warren wrote: “Self-regulation? That’s a recipe for disaster.”
Why the Superintelligence Accord Matters in Latin America
These companies’ AI tools are used across Latin America. The accord’s checks apply “during training and deployment”, in any country — but nobody can enforce them. Brazil takes another road. Its Senate passed PL 2338/2023, a bill with binding, risk-based AI rules, in December 2024. It still waits for a committee report in the Chamber of Deputies, the lower house.
Yet Brazil also wants data centres. On 15 September President Lula signed the Redata law, which suspends import duty and other federal taxes on data-centre equipment. And Trump warned on Tuesday that if US opposition continues, “they’ll be forced to go overseas or other locations”, AP reported. That puts Brazil’s power and water in play.
What to Watch Next
Trump said he will name an “AI czar”, a top adviser on AI, within three or four days, CBS News reported. The science adviser must propose a legal definition of SI by 28 November, after the midterms. Nothing is binding until laws or regulations follow — which the accord says “may make sense” one day.
Frequently Asked Questions
What is the White House Accord on Superintelligence?
It is a one-page voluntary pledge signed on 29 September 2026 by Donald Trump and six tech leaders. It asks companies building the most powerful AI models to use internal controls, an internal team, an outside auditor and a board committee.
Is the superintelligence accord legally binding?
No. The text says companies “should” take its four steps and that it “may make sense” to turn them into law later. Trump called it “morally binding”.
Has AI officially been renamed SI?
Only inside the US executive branch. An executive order of 29 September 2026 tells federal agencies to say “Super Intelligence” and “SI”. US law still uses “artificial intelligence”.
Who signed the accord?
Donald Trump, Sundar Pichai (Google), Dario Amodei (Anthropic), Mark Zuckerberg (Meta), Greg Brockman (OpenAI), Elon Musk (xAI) and Jensen Huang (Nvidia). Microsoft and Amazon, whose leaders attended the lunch, did not sign.
Sources: Truth Social post by Donald J. Trump (29 September 2026, consulted 30 September 2026), White House executive order and Ratepayer Protection Pledge, transcript of the White House press appearance (Fox LiveNOW, 29 September 2026), AP, NPR, Reuters, Axios, CBS News, CNN, ABC News, Nextgov, Al Jazeera, Meta, GovTech, NACo, Alabama Reporter, Washington Examiner, Common Dreams, Brazil’s Chamber of Deputies and Poder360 (all consulted 30 September 2026).
