Musk and Zuckerberg Kill Trump AI Safety Order
Key insights
- Musk, Zuckerberg, and Sacks successfully lobbied Trump to shelve a voluntary 90-day pre-release AI review mechanism before it was made public.
- The trio worked through allies at the National Economic Council and the vice president's office, bypassing standard regulatory channels entirely.
- The rejected order was voluntary and limited in scope, making its cancellation a significant signal about the ceiling for any US AI oversight.
Why this matters
The episode establishes a precedent that even voluntary, non-binding AI safety frameworks can be neutralized by direct billionaire lobbying before reaching public debate, compressing the window for practitioner or civil society input to zero. For founders building on frontier models, the absence of any pre-release review process means no structured signal about which capabilities regulators or governments might eventually flag as problematic, increasing long-term policy risk. Technical leaders at safety-focused labs now face a coordination problem: the informal power structure shaping US AI policy is explicitly accelerationist and has demonstrated it can act faster than any formal rulemaking process.
Summary
Three phone calls placed between Wednesday night and Thursday morning were enough to bury a White House executive order that would have required voluntary pre-release review of frontier AI models. Elon Musk, Mark Zuckerberg, and David Sacks each called Trump directly, framing the draft order as doomer regulation incompatible with US competitiveness. Their appeals landed with accelerationist allies inside the National Economic Council and the vice president's office, who helped kill the proposal before it ever went public.
The shelved order would have established a 90-day voluntary review window for frontier models before public release -- a lightweight mechanism by any regulatory standard.
Essentially: (Musk, Zuckerberg, Sacks) exercised a direct veto over federal AI safety policy with no formal authority to do so.
- The order never reached public comment or congressional review before being scrapped.
- The voluntary nature of the proposed mechanism makes its rejection a signal about appetite for even the softest oversight.
- No equivalent coalition of safety-focused researchers or civil society groups had comparable access to reverse the outcome.
The effective locus of US AI governance has shifted from regulatory agencies and elected representatives to a small group of technology executives with direct presidential access.
Potential risks and opportunities
Risks
- AI safety researchers at labs like Anthropic and OpenAI face a weakened internal bargaining position when pushing for pre-deployment review, as the White House has now signaled that even voluntary frameworks are politically untenable
- Congressional efforts to legislate AI oversight gain a harder path: opponents can now point to the executive branch's own retreat as evidence that the policy community views such mechanisms as overreach
- Allied governments coordinating with the US on AI governance frameworks -- including the EU AI Office and the UK AI Safety Institute -- may accelerate divergence from US standards, creating compliance fragmentation for frontier model developers operating across jurisdictions
Opportunities
- International AI safety bodies (UK AI Safety Institute, EU AI Office) gain credibility and talent recruitment leverage as credible alternatives to US-based governance infrastructure, attracting researchers who want structured oversight environments
- Enterprise compliance vendors and third-party AI auditors (Credo AI, Holistic AI) can position voluntary pre-release audit services as a market alternative to the shelved government framework, targeting regulated-industry buyers in finance and healthcare
- Frontier labs with existing internal safety review processes (Anthropic, Google DeepMind) gain competitive differentiation with enterprise buyers who require documented pre-release evaluation, since no government baseline now exists to commoditize that signal
What we don't know yet
- Whether any written draft of the shelved executive order has been preserved or leaked, and what specific model capabilities triggered the review mechanism
- Which officials at the National Economic Council and the vice president's office served as the internal allies who advanced the objections from Musk, Zuckerberg, and Sacks
- Whether the 90-day voluntary review concept will resurface in modified form under a different framing, or has been effectively removed from the White House policy agenda through 2026
Originally reported by semafor.com
Read the original article →Original headline: Musk, Zuckerberg, and Sacks Secretly Called Trump to Kill AI Executive Order — White House Caves to Tech Billionaires Overnight