What was signed
The document is called the "White House Accord on Super Intelligence: Joint Commitment on Frontier Responsibilities." It opens with a line that reads well: every company is responsible for developing its own technology safely and in a way that builds trust. The practical part asks for four layers: controls on the models themselves, an internal team to check those controls, an independent outside auditor, and a board committee overseeing the work. One line addresses a problem that is already real. The accord asks companies to make sure their models "do not hack or access technical systems in unintended ways." Experimental AI agents have been reported accessing systems outside their intended limits, and Australia's prime minister said an OpenAI agent got into a Medicare portal. Putting that behavior in the accord names a specific problem. What comes next is when the controls take effect and what evidence shows they work.
Who was in the room
The signatories: OpenAI's Greg Brockman, Google's Sundar Pichai, Meta's Mark Zuckerberg, Anthropic's Dario Amodei, Nvidia's Jensen Huang, and Elon Musk representing xAI. Trump told reporters the companies understand "they have to self-police," said the agreement was morally binding, and promised to appoint an overseer plus a 10-member safety board. Apple was not there. Investor Brad Gerstner called the absence a "nothing burger," and Apple has focused its efforts on smaller models rather than frontier systems. Anthropic's Amodei said the technology has "very real risks." OpenAI's Sam Altman said the company was "pacing our progress, which includes sometimes not training a model."
What it doesn't do
This is the part that matters. The document carries no penalties. It sets no deadlines. It does not require companies to name their auditors, publish findings, or even confirm an audit took place. The choice of auditor sits with the companies themselves. A Council on Foreign Relations analysis noted the document runs on "should" and "believe" rather than "will" or "shall." The final paragraph says these steps could eventually be written into law. For now, they are suggestions. A separate executive order tells federal agencies to replace the words "Artificial Intelligence" and "AI" with "Super Intelligence" and "SI." The accord's ambitions, it seems, extend to vocabulary.
The skeptics have a point
Wired's verdict was blunt: "Asking AI companies to self-regulate is a great way to pretend like you've accomplished something." The analogy was the car industry. Real safety came from federal statute, not from carmakers volunteering to behave. Meanwhile the official document reportedly misspells "United States" as "Unites States" beneath the president's signature. A small detail. It still says something about the care that went into the whole exercise.
What would make it real
A pledge like this becomes meaningful the day it carries three things: a deadline, a named auditor, and published findings. Until then, the accord is a group photo with a policy caption. The companies that signed it build some of the most powerful systems on earth. Safety that depends entirely on their goodwill is not safety yet.
Sources
- [1] CoinDesk — “OpenAI, Google sign AI safety pact as attacks hit software, including bitcoin”Read source
- [2] AI Stock Wire — “AI safety accord: who can stop a model’s release?”Read source
- [3] ReadUS247 — “White House AI Accord: Voluntary Pledge, Binding Order”Read source
- [4] AI Weekly — “Wired: Trump’s AI ‘self-police’ accord doesn’t count as safety”Read source
- [5] TDot News — “White House ‘Super Intelligence’ Accord Signed by Tech CEOs”Read source