In the days before the White House meeting, the industry’s safety argument was already splitting. OpenAI had paused training and release of an advanced model after it failed its safety standards, while Anthropic’s Dario Amodei had urged leading labs to slow down. President Donald Trump took the opposite view: the United States could not afford to restrain a technology he said would be “bigger than the industrial revolution.”¹

Tuesday’s meeting brought executives including Amodei, Meta’s Mark Zuckerberg, Google’s Sundar Pichai, OpenAI’s Greg Brockman, xAI’s Elon Musk and Nvidia’s Jensen Huang to the White House. Trump also pushed his preferred rebrand—“super intelligence,” or SI—and announced a voluntary industry accord, calling it “morally binding.”²

The agreement places the first line of defense inside the companies themselves: monitoring frontier models for cyber, biological and chemical risks; an internal team to check those controls; outside evaluators; and a board committee to receive the findings. Signatories also promise to meet regularly to set safety standards and best practices. The document concedes that these steps might eventually be written into law, but does not require it.²

For Trump and the deal’s corporate backers, that flexibility is the point. Trump said firms “have to self-police,” while Zuckerberg described the industry-wide commitment to internal controls and layered audits as a “significant positive step” that should build public confidence.Mark Zuckerberg (finkd)³ He also called the accord a start rather than the final word on AI safety.⁴

Skeptics argue that the structure merely formalizes the status quo. The companies choose and pay the evaluators, define what it means for systems to behave “as intended,” and face no stated penalty for failing the pledge. One critique called the arrangement a White House endorsement of “self-regulation” that offers neither public transparency nor independent enforcement.⁵

That gap is now the political fault line: industry leaders hail a common baseline without slowing the U.S. race against China; critics and lawmakers say the firms warning that safeguards are lagging cannot be left to write—and grade—their own rules.⁴