President Donald Trump hosted the leaders of America's biggest technology and artificial intelligence companies at the White House on Tuesday, September 29, 2026. After the lunch, Trump said the executives had signed a voluntary agreement to police the safety of their most advanced AI systems, which he described as "morally binding". There is no new law or regulation: the commitments are self-imposed, and Trump said the industry should practise "tremendous self-regulation".
Who attended the White House AI lunch?
Axios, which obtained the RSVP list, reported that 31 people confirmed attendance. The lunch was co-hosted by House Speaker Mike Johnson. Guests included:
- Jensen Huang, Nvidia
- Sundar Pichai, Google
- Satya Nadella, Microsoft
- Jeff Bezos, Amazon founder
- Mark Zuckerberg, Meta
- Elon Musk, Tesla, SpaceX and xAI
- Dario Amodei, Anthropic
- Greg Brockman, OpenAI
- Lisa Su, AMD
- Alex Karp and Shyam Sankar, Palantir
- Hock Tan, Broadcom
- Sanjay Mehrotra, Micron
- Nikesh Arora, Palo Alto Networks
- Bill McDermott, ServiceNow
- Investors David Sacks, Brad Gerstner and Chamath Palihapitiya
Administration officials including Chief of Staff Susie Wiles, Treasury Secretary Scott Bessent and Commerce Secretary Howard Lutnick also attended. According to Axios, OpenAI CEO Sam Altman did not attend because he was at his company's annual conference in San Francisco; OpenAI was represented by its president, Greg Brockman.
What is the White House AI accord?
The White House released the text of the agreement after the meeting. Forbes reporter Sara Dorn, who published its contents, and other outlets reported that the companies committed to four steps:
- Put internal controls in place so their models meet appropriate cybersecurity standards.
- Give an internal team the authority to check that those controls, monitoring and detection systems work as intended.
- Partner with an external organisation to assess their models independently.
- Designate an independent committee of their board of directors to receive those assessments and make sure problems are fixed.
Reports list the signatories as Anthropic, OpenAI, Google, Meta, xAI and Nvidia, alongside President Trump. The accord is voluntary. Speaker Johnson called it "voluntary" and Trump told reporters it was "morally" binding rather than legally binding. ABC News reported Trump described the document as a "constitution" for the industry to police itself.
Why the meeting happened now
The lunch came amid growing concern in Washington and Silicon Valley about the pace of AI development. NBC News reported that worries have intensified as cases of attacks orchestrated by AI agents mount and prominent researchers warn about serious risks. Reuters, in reporting carried by US News, said the meeting followed disclosures by AI companies of agents taking actions they had not been instructed to take. The administration has so far favoured light-touch oversight, and the accord continues that approach.
"Super intelligence" and America.gov
Trump also said he would sign an order making "super intelligence" the official US government term for AI, an idea he first floated at the United Nations General Assembly the week before, arguing that the word "artificial" makes the technology sound fake. The administration also launched America.gov, a new AI-powered government website.
What it means
For the companies, the accord formalises safety practices several of them already describe publicly, such as third-party model evaluations, without creating new legal obligations. Critics argue voluntary pledges are hard to enforce; supporters say they allow faster innovation than regulation would. What happens next depends on whether Congress acts. For now, AI oversight in the United States rests largely on companies keeping their own promises.
Why this matters beyond the US
AI rules differ sharply between countries. The European Union's AI Act imposes binding obligations on high-risk AI systems, while the US has so far relied on voluntary commitments. If you are considering working in tech abroad, these differences affect how companies build and deploy AI in each market, and which skills are in demand.
