A Handshake, Not Yet a Law
Yesterday, in the East Room of the White House, President Trump signed a voluntary AI safety accord alongside six top technology executives — a moment months in the making, arriving amid rising public alarm over how fast AI capability is advancing. The document, formally titled the "White House Accord on Super Intelligence," runs just over 300 words, creates no regulator and no penalties, and rests entirely on what Trump himself called being "morally binding."
Six Signatures, No Enforcement
Joining Trump's signature were Google's Sundar Pichai, xAI's Elon Musk, Anthropic's Dario Amodei, Meta's Mark Zuckerberg, OpenAI president Greg Brockman, and Nvidia's Jensen Huang. Notably absent was OpenAI CEO Sam Altman, who was in San Francisco for his company's annual developer conference. The accord lays out four layers of voluntary controls covering internal monitoring for cyber, biological, and chemical misuse risks, alongside a commitment that, in the document's own words, these steps "may one day" be codified into actual law and regulation — an acknowledgment that what was signed yesterday is a starting point, not a finish line.
The moment arrives directly on the heels of Anthropic CEO Dario Amodei's widely discussed essay earlier this month calling for a deliberate industry-wide slowdown in AI development, and weeks of escalating public debate over whether the technology's capabilities are outrunning the world's ability to keep it safe. Trump, who has previously dismissed AI safety warnings as a "hoax," struck a notably different tone yesterday, saying he was "seeing tremendous self-policing" across the industry. He also announced the administration has officially renamed "artificial intelligence" to "Super Intelligence" in government terminology, and said a decision on a dedicated federal "AI czar" is expected within days. Amodei, for his part, told reporters the technology carries "very real risks," but said the meeting itself had eased some of his own concerns.
Voluntary industry pledges have a long, mixed history in American technology policy, and this one arrives with the same fundamental tension as every predecessor: a handshake among competitors, backed by nothing but their own word, in an industry racing itself as much as any rival nation. Whether yesterday's accord becomes the first real step toward binding AI regulation, or simply another photo opportunity destined to be measured against what actually followed it, will depend far less on the 300 words that were signed than on what these same companies choose to do next.