Recent/upcoming developments… President Trump postponed a planned AI/cybersecurity executive order (EO) hours before a scheduled signing ceremony, saying the draft could undermine U.S. AI leadership and competitiveness with China.  Reporting suggests the delay was driven by concerns that even a nominally voluntary AI review framework could become a de facto regulatory bottleneck for frontier model releases.

* The draft EO reportedly would have created a voluntary framework for AI developers to give the federal government early access to advanced/frontier models before public release (up to 90 days in advance) so agencies could assess dangerous capabilities, identify vulnerabilities, and prepare cybersecurity defenses.  It also reportedly included provisions related to securing Pentagon and civilian federal systems, cyber hiring, threat sharing, and the use of AI tools.

* Following the postponement, Trump told reporters he “didn’t like certain aspects” of the order and did not want to “get in the way” of the U.S. lead over China, adding that the order “could have been a blocker.”  Reporting also indicates that David Sacks and tech leaders, including Elon Musk and Mark Zuckerberg, raised last-minute concerns that the review process could slow AI development or result in mandatory preclearance, even though it was voluntary on paper.

Our outlook… The federal AI regulation debate is increasingly centered on whether frontier models should be subject to some form of pre-release government review, with policymakers roughly divided among three camps: mandatory-review (Sens. Blumenthal (D-CT) and Hawley (R-MO)), voluntary-review (Rep. Obernolte (R-CA), Sen. Blackburn (R-TN)), and no-review (Sacks, Sen. Cruz (R-TX)).  While the Trump administration’s default posture has long been the latter, recent concerns over cyber/national security risks tied to advanced models (Mythos) appeared to have pushed the White House toward a targeted voluntary-review framework, with some officials (NEC Director Hasset) even signaling openness to a more formal standard.  Trump’s abrupt decision to postpone the EO underscores the internal tension between those concerns and the administration’s desire to avoid slowing U.S. AI development.  It also suggests the White House remains highly responsive to industry arguments that even voluntary review could become problematic.  We preliminarily expect the EO to reemerge, though likely in a narrower, lighter-touch form focused on best practices, standards, and/or agency coordination.  In Congress, that posture would make the White House more likely to back Cruz/Blackburn/House GOP-style light-touch frameworks than Hawley-Blumenthal-style mandatory evaluation proposals.

* In contrast, state activity remains mostly focused on consumer-facing AI rules rather than pre-release model review.  State frameworks, predominantly, emphasize transparency, opt-out or human-review rights, documentation obligations, and nondiscrimination protections (e.g., CA, CO, CT, TX, UT).  Where states are moving closer to scrutinizing underlying frontier models (e.g., CA, NY, IL), the policy framework is still not “review and consent,” but the creation of a mandatory safety/transparency infrastructure for certain models (i.e., published safety frameworks, catastrophic-risk assessments, transparency reports, critical-incident reporting, third-party evaluation or audits, whistleblower protections, and AG/state oversight).

Watch for these developments… We continue to watch for further developments regarding the sorting of lawmakers into the various camps on the review/consent issue.  For Republicans, the key marker is whether the two camps that make up the bulk of the party (voluntary and no-review) can settle on a unified approach, as this would increase the odds of congressional action this year (even if on a partisan basis).  For Democrats, we are watching for indications that lawmakers (who generally prefer more aggressive review requirements than their Republican counterparts) are willing to accept a middle-ground framework short of mandatory pre-release review (which would likely need to be paired with robust consumer-facing protections to secure their support).  If Democrats move in this direction, it could open the door to bipartisan compromise (assuming Republicans have coalesced around an approach that includes more than simply no review).