AI Policy Wiki
Dashboard

California Just Played the One AI Card the Feds Can't Block (Clearwater, March 2026)

high confidence · updated 2026-06-06

Andrew Clearwater's analysis of California EO N-5-26 (Newsom, March 30, 2026). Frames the order as Newsom 'finding the legal seam in the federal preemption wall' — exploiting Section 8 of Trump's December 2025 federal preemption EO, which carves out state procurement from preemption scope.

"California Just Played the One AI Card the Feds Can't Block" is an essay by Andrew Clearwater, published March 31, 2026 (Source: https://andrewclearwater.substack.com/p/california-just-played-the-one-ai). It analyzes California EO N-5-26, signed by Governor Newsom on March 30, 2026, and argues that the order is structured to operate through state procurement, a channel Clearwater reads as carved out of the federal AI-preemption regime.

Summary of argument

Clearwater argues that EO N-5-26 should not be read as "another AI executive order," but as Newsom "finding the legal seam in the federal preemption wall." On Clearwater's reading, Section 8 of Trump's December 2025 federal preemption order explicitly carves out state procurement and governmental use of AI from the scope of preemption. Because N-5-26 is, in Clearwater's framing, entirely about procurement and does not regulate what AI companies can build, it operates as a condition on access to California contracts rather than a direct rule on AI development: in his characterization, if a vendor wants California's money, the order sets out what it must demonstrate about its safety practices. Clearwater frames this as the purchasing power of the world's fourth-largest economy used as a governance tool through a channel that federal preemption cannot reach.

Key claims

On the content of the order, Clearwater describes EO N-5-26 as directing the California Department of Technology (CDT), the Department of General Services (DGS), and other agencies to develop recommendations within 120 days, giving a late-July 2026 deadline. He anticipates that the resulting framework will likely require AI vendors seeking state contracts to attest to and explain their policies across three risk categories: illegal-content prevention (CSAM, NCII); bias governance; and civil-rights protections (free speech, voting, autonomy, anti-discrimination, anti-surveillance). He notes that CDT is to develop what he describes as the first state-level watermarking guidance for AI-generated images and manipulated video. Clearwater emphasizes that the order sets no certification standards yet (those are to come out of the 120-day process), imposes no penalties beyond existing procurement law, and adds no model-level technical requirements.

Clearwater places the order in a federal sequence: the Biden administration's EO 14110 (a broad AI safety order) in October 2023; Trump's revocation of EO 14110 on January 20, 2025; Trump's EO 14179, "Removing Barriers to American Leadership in AI," on January 23, 2025, which he describes as flipping the federal posture from oversight to deregulation; the July 2025 AI Action Plan, under which states regulating AI too aggressively risk losing federal funding; Newsom's signing of SB 53 in September 2025, which Clearwater calls the first enforceable US frontier AI safety statute; and the December 11, 2025 federal preemption EO, which established a DOJ AI Litigation Task Force and, on Clearwater's reading, carves out state procurement. N-5-26, signed March 30, 2026, is the procurement-focused move in this sequence.

Clearwater argues the order matters even to companies that do not sell to California. He contends the attestation model becomes a template: when the fourth-largest economy in the world standardizes vendor questions, in his view those questions become the market's questions. He projects that other states will follow, framing this as a California Effect amplified through procurement, and argues that procurement requirements radiate inward because companies do not build a parallel governance program for one customer but build their single governance program around such requirements. Clearwater summarizes his outlook in the essay's closing passage:

Procurement-driven governance is the operating model for the foreseeable future. The federal regulatory ceiling isn't coming. What's coming is a world where your biggest customers require you to demonstrate responsible AI practices as a condition of doing business.

Recommendations

The essay closes with six steps Clearwater prescribes: (1) read the actual order (three pages) and map current governance documentation against the three attestation categories; (2) track CDT and DGS recommendations, noting that the 120-day clock started March 30 and that late July 2026 is when the policy becomes operational; (3) watch the federal preemption fight, because the procurement carve-out lives in an executive order rather than a statute and could change; (4) evaluate vendor contracts; (5) build documentation now; and (6) not wait for final rules.

Provenance

The essay advances Clearwater's framing of procurement-driven AI governance as the operating model toward which he argues US AI governance is moving, identifies Section 8 of the December 2025 preemption order as the legal mechanism he sees enabling state action under federal preemption pressure, and forecasts California Effect amplification through procurement. As an essay, its claims are positions rather than established findings, including its reading of the Section 8 carve-out and its projection that other states will follow.

Relationships