Trump and tech CEOs sign voluntary ‘Super Intelligence’ safety accord
After a White House lunch Tuesday, President Trump announced that leaders of Anthropic, Google, Meta, OpenAI, Nvidia, and xAI signed a voluntary Joint Commitment on Frontier Responsibilities for AI safeguards, while an executive order directed agencies to say “Super Intelligence” instead of “artificial intelligence.”
AI-assisted · US Brief desk · Sources listed below
What happened
Trump told reporters outside the West Wing he was seeing “tremendous self-policing,” AP reported. He posted the Joint Commitment on Frontier Responsibilities to Truth Social; signatories named across AP, BBC, and Al Jazeera include Anthropic CEO Dario Amodei, Google CEO Sundar Pichai, Meta CEO Mark Zuckerberg, OpenAI President Greg Brockman, Nvidia CEO Jensen Huang, and xAI’s Elon Musk (with xAI described as now part of SpaceX in some accounts). House Speaker Mike Johnson also joined the gathering.
Why it matters
The White House is answering rogue-agent scares and midterm data-center backlash with voluntary industry promises and a nomenclature shift — not statutory guardrails — while competing with China and rejecting international AI oversight schemes.
What’s next
Watch whether Congress tries to write the four layers into law, how companies name auditors, and whether agencies actually purge “AI” from public documents.
More context
President Donald Trump said Tuesday that he and leading AI executives signed a voluntary accord on frontier-model safeguards after a White House lunch — and separately ordered federal agencies to call the technology “Super Intelligence,” not artificial intelligence.
What the accord says: Outlets describe four voluntary layers — internal monitoring during training and deployment; empowering internal teams to remediate issues; partnering with independent external auditors; and board-level oversight. Trump called it “almost like a constitution” and “morally binding.” The Guardian and Al Jazeera note it appears to carry no enforcement mechanism, lets companies choose auditors and boards, and does not require publishing evaluation results; the text says it “may make sense” to codify measures later.
Super Intelligence order: Trump signed an executive order directing departments and agencies to use “Super Intelligence” and “SI” in official correspondence, websites, reports, and non-statutory documents, and to stop acknowledging “Artificial Intelligence” or “AI,” BBC and The Guardian reported — echoing language he used at the U.N. last week.
Data centers: Facing bipartisan local opposition to AI data centers, Trump said companies would make “massive contributions to communities” and predicted the facilities would become popular; Huang called them “super intelligence factories,” BBC reported. A Marist/NPR/PBS poll cited by BBC found 65% of registered voters opposed data centers in their areas.
Pushback: Critics quoted by BBC, Guardian, and Al Jazeera called the pact vague self-regulation. Geoffrey Hinton told CBC Trump “doesn’t really understand AI.” The meeting came a day after OpenAI said it shelved GPT-6.1 Astra over safety concerns.
Uncertainty: Guest lists and who formally signed versus merely attended differ slightly by outlet. The accord’s legal effect is, by design, unclear. US Brief has not independently reviewed the full Truth Social PDF beyond secondary descriptions.
4 listed sources
- Associated Press via CBC News, Sep 29, 2026: Trump says top tech firms have signed accord to 'self-police' AI development
- BBC News (Osmond Chia), Sep 30, 2026: Three takeaways from Trump's 'Super Intelligence' summit
- The Guardian (Nick Robins-Early), Sep 29, 2026: Trump announces vague ‘morally binding’ AI deal among tech CEOs
- Al Jazeera (John Power), Sep 29, 2026: Trump, tech bosses sign voluntary pact pledging ‘robust’ AI safeguards
References listed by US Brief; a source count is not a verification score.
Editorial sourcing notes
Update of pre-meeting draft. Post-meeting sourcing: AP via CBC/The Gazette, Sep 29–30, 2026 — self-police quotes, signatory list (Amodei, Pichai, Zuckerberg, Brockman, Huang, Musk), Truth Social post of accord, data-center community pledges. BBC (Osmond Chia), Sep 30, 2026 — three takeaways, Super Intelligence EO text, four voluntary layers, expert skepticism. The Guardian (Nick Robins-Early), Sep 29, 2026 — “morally binding,” no enforcement, companies pick auditors. Al Jazeera (John Power), Sep 29, 2026 — Joint Commitment details, Sacks/Graylin/Walsh/Krueger reactions, OpenAI Astra pause context. Guest lists vary slightly across outlets (some mention Bezos/Nadella attending lunch without signing).
Like US Brief? You can support it with a tip.