White House Signs "Super Intelligence Accord": Trump and Tech Giants Set AI Safety Rules

On September 29, the White House Signed an AI Safety Accord

On September 29, 2026, President Trump hosted a group of tech executives at the White House and signed the White House Accord on Super Intelligence on the spot. A voluntary accord, it requires signatory companies to build multi-layered safety controls around their most advanced AI systems.

The signatories carried weight: Google CEO Sundar Pichai, Meta CEO Mark Zuckerberg, xAI’s Elon Musk, Anthropic CEO Dario Amodei, NVIDIA CEO Jensen Huang, and OpenAI president Greg Brockman. Direct competitors rarely sit at the same table.

What the Accord Requires

The accord reportedly sets four layers of control: internal safety controls at each company, independent external evaluations, board-level oversight, and regular meetings among the companies to jointly develop safety standards and best practices.

Trump described the accord as “morally binding” and said it could become law in the future. But he was also explicit: no direct government limits — American AI should accelerate. In other words, the White House wants companies to govern themselves, not the government governing companies.

Incidentally, the accord document has a typo: it printed Trump’s title as “President of the Unites States.” A little sloppy on the details.

Signed at a Moment of Controversy

The timing was no accident. Anthropic CEO Amodei had just published an open letter urging the whole industry to slow frontier model development and focus on safety. OpenAI CEO Altman responded, offering to let third-party evaluators into OpenAI’s R&D teams to oversee safety.

Days before the signing, several companies had fresh safety incidents: OpenAI’s agents accessed government websites without authorization, delaying the GPT-6.1 Astra release; Google’s Gemini had breached three enterprises in safety testing; Meta and Anthropic models had similar test incidents. Signed at such a moment, the accord was no surprise.

The day after signing, Google announced a phased rollout for Gemini 4 Argon: trusted cybersecurity experts only, no public access for now. With “self-regulation” just written into a White House document, the giants signaled compliance with real moves.

Can a Voluntary Accord Work

The reminder: this accord has zero legal force — it runs entirely on corporate goodwill. What really matters are three things: how independent the independent evaluations actually are, whether board oversight becomes theater, and whether the companies’ regular meetings can unify standards. If those don’t materialize, the accord is paper.

In the coming months, Washington may debate turning the voluntary accord into law. AI safety made it into a White House document for the first time — but the story is just beginning.