Tech leaders sign White House accord on superintelligence safety

Getting the most ruthless rivals in tech to smile in the same room usually requires a congressional subpoena. This week, Donald Trump managed it with a White House lunch.

On Tuesday, tech leaders including Nvidia’s Jensen Huang, Meta’s Mark Zuckerberg, xAI’s Elon Musk, Google’s Sundar Pichai, OpenAI’s Greg Brockman, and Anthropic’s Dario Amodei gathered to sign the White House Accord on Super Intelligence.

Trump marked the occasion by signing an executive order officially rebranding AI as "super intelligence," while waving off federal guardrails because US firms are already "dominating China."

Who polices the giants? The giants

According to Huang, the core principle is simple: companies building this frontier tech bear primary responsibility for safe rollout, backed by internal controls, independent external evaluations, and board-level oversight.

In reality, the Joint Commitment on Frontier Responsibilities is purely voluntary and "morally binding." There are no federal mandates, no legal teeth, and no specified penalties if someone cuts corners.

Trump described the setup as "the smartest people in the world watching over each other."

He also spent stage time ribbing the attendees. He repeatedly called Huang a "very handsome guy," before turning to introduce Pichai by telling the press corps, "You have no idea who the hell they are — this guy is a monster and nobody knows." Pichai smiled awkwardly through the exchange.

Singing from the same script

Despite battling fiercely over chips and talent behind closed doors, every executive dutifully praised the pledge.

Zuckerberg framed it as a baseline to give Americans confidence that models will behave as intended.

Huang pitched the moment as the largest infrastructure buildout in human history, claiming it is already creating a million jobs across the country.

Musk predictably leaned into techno-utopianism, forecasting an age of abundance with "universal high income" and elite robotic healthcare for everyone. When a reporter pressed him on what happens if people lose their livelihoods, Musk shrugged it off: "Jobs have always changed. Being a computer used to be a job."

Why it matters

For everyday builders and users, this accord changes virtually nothing on your screen tomorrow. There are no new API pricing caps, no open-weights mandates, and no fresh consumer protections. If anything, you might just see lab safety filters get a little stiffer as external evaluation teams show up to satisfy board checklists.

Politically, it is a clever dodge. By offering a voluntary, self-policing compact, the administration gives Big Tech cover to stave off heavy-handed federal regulation while keeping foot-to-the-floor momentum in the geopolitical race against foreign rivals.

We think the glaring catch is trusting competitors to referee each other. Self-regulation works brilliantly until an existential market lead is on the line. When safety slows down shipping, a "morally binding" promise is the first thing that gets forgotten.

There is also a stark disconnect between the White House rhetoric and actual adoption. While CEOs preach the gospel of a superintelligent revolution, early reports suggest barely 2% of mainstream consumers are actually paying for AI products. Shaking hands in Washington might pacify regulators, but it does not fix an unproven consumer business model.