President Trump scrapped a planned AI safety executive order on May 21, 2026 — hours before it was scheduled to be signed — after Elon Musk, Mark Zuckerberg, and David Sacks intervened, calling him directly to warn that the proposed review framework would slow American AI development.
Scrapped Hours Before Signing
The signing had been scheduled at the White House, but the ceremony was pulled at the last minute after the direct calls. According to Axios, the three technology executives argued that a formal pre-release review framework for frontier AI models would burden the leading AI labs and hand an advantage to rivals they claim are moving faster.
Who Called and Why
Musk, Zuckerberg, and Sacks each have direct lines to the administration on AI policy, and each urged the White House to keep the industry free of what they described as heavy-handed review requirements.
- Elon Musk argued that federal review would slow the pace of model releases
- Mark Zuckerberg warned that testing requirements would tilt the field toward incumbents with less to lose
- David Sacks, the administration’s AI and crypto point person, echoed concerns that the framework would get in the way of US AI leadership
What the Order Would Have Done
The cancelled order would have established a government framework to evaluate frontier AI models for safety and security ahead of public release — the kind of pre-release review that safety advocates have pushed for but that the industry has resisted as an innovation tax. The framework would have applied to the most capable systems produced by leading US labs.
Reaction
The scramble drew immediate criticism from safety advocates, who said the pattern made clear that the administration’s AI policy bends to direct lobbying from the largest companies. Supporters of the decision framed it as the right call to keep the US AI industry dominant.
What Happens Next
With the order off the table, no binding federal standard now covers pre-release frontier AI safety. The episode signaled that, under this White House, formal AI review frameworks face steep resistance — and that the executive suite still carries more weight in AI policy than any draft rule.