On Tuesday, in a room in Washington D.C., seven men signed a document that Donald Trump compared to a constitution. The signatories were not lawmakers but the people who currently shape the digital lives of billions: Sundar Pichai of Google, Elon Musk of SpaceX, Mark Zuckerberg of Meta, Jensen Huang of Nvidia, Dario Amodei of Anthropic, and Greg Brockman representing OpenAI.

What they agreed to was described as "morally binding" rather than legally enforceable, a distinction that will matter enormously in the months ahead. The pact commits each company to testing its own AI systems, fixing problems quickly, and hiring independent auditors to confirm that the technology behaves as intended and does not quietly slip past the safeguards built around it. Trump said a new oversight board would be created, though he gave no details about who might sit on it. For Zuckerberg, the agreement was reassurance for ordinary users; for Amodei, it was, in his words, a way to compete without losing control of the technology altogether.

Not everyone was persuaded. Jose Miguelito Enriquez, an AI governance researcher at Nanyang Technological University in Singapore, noted that the accord signals growing agreement among companies but does little to guarantee real accountability, since the same firms that write the rules are also the ones grading their own homework. That tension — between Silicon Valley's instinct to move fast and a public increasingly asking who watches the watchers — sits at the heart of the agreement.