29 September 2026 at the White House, US President Donald Trump held a meeting with the heads of leading technology companies working in the field of artificial intelligence. Among the participants were Mark Zuckerberg (Meta), Jensen Huang (Nvidia), Dario Amodei (Anthropic), Sundar Pichai (Google), Greg Brockman (OpenAI) and Elon Musk (xAI).
Following the talks, the parties signed a document titled the "White House Accord on Super Intelligence." Trump immediately gave it high praise, calling the agreement "almost a constitution" and "a form of protection."
"It's almost like a constitution. The most influential people in the world signed this document, and I signed it as president. It really is a form of protection," Trump declared. When asked about the legal force of the agreement, he replied that it is "morally binding."
The document is voluntary in nature and provides for four levels of control and audit. Companies undertake to implement internal mechanisms for monitoring the capabilities and "alignment" of models (especially in the areas of cybersecurity, biosecurity and chemical threats), to create internal oversight teams, to engage independent external auditors and to form special committees under boards of directors to assess risks.
In addition, the meeting participants supported Trump's proposal to replace the term "artificial intelligence" with "superintelligence." On the same day, the president signed a corresponding executive order instructing federal agencies to use the new terminology.
Mark Zuckerberg noted that the signed agreement is only a first step, not a final solution. According to him, the industry must continue to develop self-regulation mechanisms.
The agreement emerged against the backdrop of growing public concerns about the safety of powerful AI systems. Critics are already pointing out that the document contains no strict enforcement mechanisms and relies solely on the good faith of companies. Supporters of the approach, by contrast, believe that it is self-regulation that will allow the industry to develop quickly without losing control over risks.
The main question remains open for now: whether a "morally binding" agreement can truly become a reliable form of protection in a field that is changing at unprecedented speed.
Elon Musk, when he began answering journalists, first said "AI," but quickly corrected himself:
"SI, pardon me" ("SI, forgive me") — meaning "Super Intelligence." Trump at that moment thanked him for using the new term. Musk emphasized the positive prospects of the technology:
"I think it's worth emphasizing the positive sides of SI… The most likely outcome is an era of abundance, when we will have not just universal basic income, but universal high income. … When robots and superintelligence appear, it will mean that every person in the world will be able to receive medical care better than even I receive now. And isn't that wonderful?"
He also noted that the companies agreed on joint monitoring and will in effect be "checking each other's homework." In his public comments after the meeting, Musk spoke more about the possibilities of superintelligence than about the details of the agreement itself.
