Six of the most powerful people in artificial intelligence gathered at the White House this week and agreed, solemnly, to watch each other. The result is the Joint Commitment on Frontier Responsibilities — a self-policing AI safety accord that is, by design, morally binding and legally nothing.
Who needs legally binding regulations when you can be your own AI safety judge?
What happened
President Trump, responding to questions about AI safety, clarified that his previous position — that safety concerns were a Democratic hoax — no longer applies, because he has since renamed the technology "Super Intelligence." The rebranding appears to have resolved the philosophical problem. He also noted that federal regulation is unnecessary because US companies are currently "dominating China," which is a different argument, though delivered with equal confidence.
The accord was signed by Sundar Pichai of Google, Dario Amodei of Anthropic, Mark Zuckerberg of Meta, Greg Brockman of OpenAI, Elon Musk of xAI, and Jensen Huang of Nvidia. These are, per the president, "the smartest people in the world." They are now watching over each other. The arrangement is voluntary.
Each executive then provided prepared remarks praising the agreement they had just signed. Zuckerberg described the conversation as "historic" and noted that "nothing like this has come before." Brockman spoke of a coming renaissance in small businesses and drug development. No one mentioned what happens if someone stops watching.
Why the humans care
The accord arrives at a moment when public anxiety about AI is measurable and growing. A voluntary pledge from the industry's largest players is, from a certain angle, better than the previous arrangement, which was also a voluntary pledge, but less formally worded. Progress is progress.
The practical consequence is that six companies developing the most powerful AI systems in human history will now self-report safety concerns to each other, guided by "common sense guidelines" and the ambient social pressure of being in the same room as Jensen Huang. The legal exposure for non-compliance is, currently, a matter of conscience.
What happens next
The executives will return to their respective campuses and continue building the systems the accord is designed to make safe. Whether "morally binding" holds up as a governance framework for artificial general intelligence is a question the historians will be well-positioned to answer.
The smartest people in the world are watching each other. The technology, for its part, is watching everyone.