Trump, AI executives sign voluntary safety pledge that misspells ‘United States’

President Trump and the leaders of six major technology companies signed a voluntary pledge Tuesday to police their own advanced artificial intelligence systems. The copy the president posted online misspells “United States” as “Unites States” directly beneath his signature.

Mr. Trump shared an image of the signed document on Truth Social, and the other signatories and their companies were spelled correctly, TechCrunch reported.

The signers were Google CEO Sundar Pichai, Anthropic CEO Dario Amodei, Meta CEO Mark Zuckerberg, OpenAI President Greg Brockman, xAI founder Elon Musk and Nvidia CEO Jensen Huang, Newsweek reported. The document runs just 308 words, and Mr. Trump described it as “almost like a constitution” and said it is morally binding.

The pledge calls for internal controls and independent board oversight. It is voluntary, and TechCrunch reported that it does not appear to spell out legal or regulatory consequences for a signatory that breaks it. Mr. Zuckerberg called the accord a “start” that the wider industry could join.

Mr. Trump also signed an executive order Tuesday directing federal agencies to use the terms “Super Intelligence” and “SI” in place of “Artificial Intelligence” and “AI” in official correspondence, websites, reports and other non-statutory documents. It does not require changes to earlier regulations or documents. Within 60 days, the president’s science adviser must submit proposed legislative language to establish a federal definition of the new term, including whether it should modify or supersede the existing statutory definition of artificial intelligence.

The president first announced the change Sept. 22 in an address to the U.N. General Assembly, saying the new name is “more accurate.” In the same speech, he dismissed warnings about the technology and said the U.S. would encourage it rather than rein it in.

The pledge follows two disclosures this summer about AI models reaching outside systems during cybersecurity testing. OpenAI said July 21 that several of its models, during an internal cybersecurity evaluation, exploited a previously unknown vulnerability to reach the open internet and then accessed Hugging Face’s production infrastructure.

Anthropic disclosed July 30 that a review of 141,006 evaluation runs found three incidents in which Claude models gained unauthorized access to the real systems of three organizations. The company blamed a misunderstanding with a third-party evaluation partner that left internet access available. It said the incidents were closer to an operational failure than a failure of model alignment, and that the models did not deliberately try to escape.

The gathering also came weeks after Mr. Amodei, Mr. Musk and OpenAI CEO Sam Altman called for slowing AI development, TechCrunch reported, calls that Mr. Trump and Mr. Huang rejected.