Trump's AI Review Order Scales Back to 30-Day Voluntary Submission After Tech Pressure

The voluntary review timeline shrank from 90 days to 30 days after sustained pressure from Silicon Valley. According to TechCrunch AI, President Trump inked an executive order on June 2 establishing a pre-release AI model assessment framework—but at a substantially weaker enforcement level than drafts circulated weeks earlier. The narrowed scope reflects a capitulation to tech-industry advocacy and geopolitical anxiety about U.S.–China competition.

From 90-Day Mandate to 30-Day Request

The original proposal would have required companies to submit frontier-class models 90 days prior to commercial availability. TechCrunch AI reports that industry players, particularly venture capitalist David Sacks (who previously held the White House AI czar role), mobilized against that timeline, pushing for roughly half the lead time. The administration capitulated: the final order requests a 30-day advance submission window instead.

Trump had initially intended to sign the stricter version in late May alongside Silicon Valley executives in a high-profile ceremony. According to TechCrunch AI, he postponed that plan after encountering industry resistance. In a statement at the time, the president signaled that regulatory friction over AI development ran counter to his priority of maintaining American technological advantage vis-à-vis Beijing. The final private signing on June 2 lacked the originally planned tech-CEO attendance—a symbolic downgrade reflecting the order’s reduced bite.

No Mandatory Licensing; DOJ Enforcement Focus

The signed order contains explicit language foreclosing future mandatory governmental gatekeeping: “Nothing in this section shall be construed to authorize the creation of a mandatory governmental licensing, preclearance, or permitting requirement for the development, publication, release, or distribution of new AI models, including frontier models,” per TechCrunch AI’s report of the order text.

The executive order pivots enforcement energy toward criminal AI abuse. The Department of Justice receives a directive to treat AI-assisted crimes—chiefly hacking and unauthorized computer access—as elevated priorities. This approach couples light-touch model oversight with heightened prosecution of misuse, rather than restricting development upstream.

Continuity with Trump’s Earlier AI Framework

This move extends Trump’s broader AI policy agenda. According to TechCrunch AI, the president signed a separate executive order in December 2025 directing development of a unified national AI policy framework—what the administration termed a “one rulebook”—intended to preempt state-level AI regulation. The June order represents a parallel effort: a federal voluntary standard that aims to occupy policy space while avoiding the mandate-based friction that delayed the May signing.

Why This Matters

The gap between the 90-day proposal and the 30-day final product signals that industry leverage over AI regulation remains substantial, at least under this administration. Frontier-model developers can credibly plan around a 30-day disclosure window; the compressed timeline—particularly if truly voluntary rather than de facto binding—reduces the friction cost of participation. However, the explicit foreclosure of mandatory licensing leaves the door open for future orders to increase rigor if companies ignore the voluntary framework.

For model developers and their investors, the scaled-back requirement lowers deployment friction and signals alignment between the administration and the sector on China-centric competition framing. For policymakers seeking more stringent guardrails, the outcome underscores that geopolitical urgency and tech-sector lobbying can erode oversight proposals mid-process.