29 September 2026 at the White House, Sam Altman and Dario Amodei stood alongside Donald Trump and other tech company leaders, signing a one-page document on "joint commitments to responsibility for frontier systems." Trump called it "almost a constitution" and "morally binding," and the companies promised internal controls, external audits, and board committees — without fines or mandatory norms.
PRESIDENT TRUMP AND MAJOR AI LEADERS SIGN A VOLUNTARY WHITE HOUSE ACCORD ON SUPER INTELLIGENCE, COMMITTING TO STRONGER SAFETY CONTROLS, OUTSIDE AUDITS AND INDEPENDENT OVERSIGHT
Previously, Altman and Amodei had publicly demanded slowing down development and introducing government rules: Amodei in an essay called for independent audits and even an antitrust exemption for discussing safety, while Altman supported the idea of a "frontier pace." Now they have signed a voluntary agreement that Trump called "self-regulation" and that echoes Biden's approach from 2023, but without real leverage.
For Amodei and other industry leaders, this is not a sudden turn but a continuation of strategy: they have always balanced between warnings about risks and the desire to maintain leadership in the race with China. Their reputation was built on public calls for responsibility, but real actions — from infrastructure investments to rejecting strict restrictions — showed the priority of speed. Under pressure from Trump, who rejects tough regulation and promotes the term "superintelligence," they chose a form that allows them to look responsible without losing control.
The agreement introduces no new restrictions, merely formalizing what companies were already doing or planning. It opens the door to future laws but leaves them to the discretion of the firms themselves. This confirms a long-standing pattern: when the threat of real regulation grows, industry leaders prefer voluntary frameworks that do not hinder development and do not require concessions to competitors.
Imagine a company that for years complains about the lack of traffic rules, and then draws the signs itself and promises to obey them — provided the police do not intervene. That is exactly what the current step looks like: public concern for safety masks the desire to preserve freedom of maneuver.
This episode shows how powerful players use political shifts to turn a potential threat into a tool for strengthening their positions. Self-regulation becomes not a compromise but a way to postpone real restrictions indefinitely.
In the end, AI leaders have proven that their public calls for control were always part of the game to keep the initiative in their own hands.
