Trump's Superintelligence Accord Gets Big Tech's Pinky Promise

Six of the most powerful AI executives on the planet flew to Washington this week, sat down in the White House, and signed a document promising to be careful with the most disruptive technology of the century. There's just one catch: the promise isn't legally binding. It's, in President Donald Trump's own words, 'morally binding.' Welcome to the superintelligence accord — part policy, part theater, and all anyone in tech can talk about right now.

The signing happened Tuesday, Sept. 29, when Trump and House Speaker Mike Johnson hosted roughly 20 tech leaders for a White House meeting on artificial intelligence. According to Reuters, six executives put their names on the one-page 'White House Accord on Superintelligence: A Joint Commitment on Frontier SI Responsibilities': Google CEO Sundar Pichai, Anthropic CEO Dario Amodei, Meta CEO Mark Zuckerberg, OpenAI President Greg Brockman, Nvidia CEO Jensen Huang, and Elon Musk of xAI.

What the accord actually requires

The document is only a page long, and its four commitments are the kind of things companies are fond of agreeing to voluntarily: robust internal controls that monitor models during training and deployment for cybersecurity, biosecurity, and chemical threats; partnership with an independent external auditor; an independent board committee to review audit reports; and regular meetings among the signatories to set standards going forward. According to reporting by the Associated Press, the accord explicitly opens the door to future legislation but imposes zero federal enforcement today.

That gap between ambition and enforcement is the whole story. Trump framed the accord as 'almost like a Constitution in a way,' a founding document for the superintelligence era, and floated creating an oversight committee of around 10 people. But a constitution nobody can enforce is, well, a strongly worded suggestion. The executives themselves already perform much of what's promised — companies already run red-teaming and safety evals — which is exactly why experts quoted in the AP coverage called the accord vague — the same week we covered OpenAI shelving a model over AI safety concerns and noted that only 'robust legal liability, regulation' actually reduces harm, as Berkeley AI policy researcher Alex Pascal put it.

Amodei, the Anthropic CEO and the most safety-hawkish voice in the room, offered the most sober note of the day: 'The technology has very real risks.' It's worth sitting with that for a second — the man whose company is racing to build frontier AI systems said so at a ceremony celebrating a voluntary pact his own company signed.

The 'Super Intelligence' rebrand and the America.gov chatbot

If the accord was the policy, the branding was the show. Trump announced he has signed a directive ordering federal documents to replace the term 'artificial intelligence' with 'super intelligence,' arguing that the technology is, quote, 'not artificial.' Love it or hate it, it's a deliberate signal that the White House wants AI treated as a nation-defining project on par with the space race — and a nickname that launched a thousand memes within the hour.

Trump also unveiled an America.gov AI chatbot designed to help people navigate government services, with Secretary of State Marco Rubio suggesting it could eventually process first-time passport applications fully online. The executives, for their part, promised to soften local opposition to data-center construction — one of the fastest-growing sources of NIMBY friction in America — by funding schools and cutting energy costs in host communities. And Trump framed everything in the language of a US-versus-China contest with 'no second place.'

Why a pinky promise isn't regulation

Here's why this matters beyond the photo op. The people profiting most from AI now get to grade their own homework, with a voluntary accord instead of enforceable rules. That is precisely the critique consumer advocates have leveled at every era of self-regulation, from social media to airlines: when the industry writes the test and takes it themselves, the passing grade is guaranteed.

None of this is to say the accord is meaningless. Getting Pichai, Zuckerberg, Amodei, Brockman, Huang, and Musk to publicly commit to external audits and board-level review creates a paper trail — and paper trails become leverage when Congress eventually writes real law. Regular meetings among rivals also matter in a field moving too fast for any single company's safety team to track alone. But a 'morally binding' promise is only as strong as the public memory of who signed it, which is why reporting like the Reuters coverage of the signing and the AP report on the accord's details will be cited the first time a signatory breaks it.

For Gen Z — the generation that will live longest under whatever superintelligence becomes — the superintelligence accord is a perfect explain-it-to-a-friend story: the AI companies signed a pinky promise to behave, the president renamed the whole field like it was a vibe shift, and whether any of it holds depends on whether anyone with actual power decides to turn the moral promise into a legal one. Watch what Congress does next. The accord opened that door; walking through it is a choice.