The Information Machine

US AI Regulation: Federal Retreat vs. State Intervention · history

Version 12

2026-06-04 08:19 UTC · 229 items

What

Trump signed an executive order on June 3 establishing a voluntary 30-day pre-release review process for frontier AI models [8][9] — a revised version of an earlier order canceled in May, with the review window reduced from 90 days and all binding requirements stripped [10]. The same day, OpenAI published a public policy agenda and a blueprint for democratic governance of frontier AI explicitly calling for a coherent federal framework over fragmented state rules [21][22]. State-level action continues independently: Illinois SB 315 is enacted law requiring mandatory third-party auditing [13], Florida's civil lawsuit and criminal investigation against OpenAI are active [19][26], and Senator Sanders' American AI Sovereign Wealth Fund Act — proposing a one-time 50% stock transfer from major AI companies to a public fund [24] — has drawn wide public attention but no reported co-sponsors.

Why it matters

The Trump EO creates the first federal touchpoint for frontier AI oversight, but its voluntary design and classified benchmarking thresholds leave enforcement questions open [8], while DOGE-driven cuts to federal security teams raise practical questions about whether the government can conduct real evaluations at all [10]. OpenAI's push for a federal framework — framing frontier AI as a national security matter [22] — puts the company in direct opposition to the accelerationist faction that just succeeded in weakening the EO.

Open questions

  • Trump's AI EO is nominally voluntary, but Zvi Mowshowitz argues labs have no practical choice but to comply, making it de facto mandatory prior restraint [8] — will courts treat it as such, and does the classified threshold structure survive transparency challenges?

  • DOGE-era cuts have reduced federal cybersecurity evaluation capacity [10] — can the government conduct substantive 30-day safety reviews, and what is the consequence if reviews are perfunctory?

  • OpenAI is explicitly advocating for a federal governance framework [22][23] — does this produce legislative coalition-building that could pass federal law preempting the current state-level patchwork?

  • Illinois SB 315 is enacted law [13] and Florida's litigation is active [19] — does the new federal EO affect either track, or do they proceed on entirely separate constitutional ground?

Narrative

American AI governance in 2026 is defined by a federal administration that has oscillated between blocking state oversight and installing a limited federal alternative, while states have built parallel accountability structures that do not depend on federal coordination. The Trump administration's preemption executive order challenged state AI laws on constitutional grounds [1][2], but multiple law firms found its authority limited absent congressional authorization [3][4][5]. An earlier AI security order was canceled in May after industry lobbying: Sacks called Trump directly without White House staff knowledge, arguing pre-deployment review would hurt the US-China race, while Musk and Zuckerberg lobbied the accelerationist faction against it [6][7]. OpenAI reportedly supported signing, with executives mid-air to Washington when the event was canceled.

Trump subsequently signed a revised AI executive order on June 3 establishing a voluntary framework for pre-release review of frontier models, with a 30-day government access window reduced from the canceled draft's 90 days [8][9]. The order explicitly disclaims creating a licensing regime and establishes no binding requirements on AI firms [10]. Analyst Zvi Mowshowitz argues it nonetheless functions as de facto mandatory because labs have no practical choice but to participate, and that classifying the benchmarking thresholds is particularly problematic since researchers at AI labs will not know whether their own models trigger the review requirement [8]. Ars Technica reports that DOGE-driven cuts to federal cybersecurity and safety teams have reduced the government's practical capacity to conduct real evaluations, framing the result as performative rather than substantive assurance [10].

At the state level, Illinois became the most stringent AI safety jurisdiction in the country when Governor Pritzker signed SB 315, requiring the largest AI firms to publish annual third-party safety test results, submit public safety plans, and report critical incidents within 72 hours [11][12][13]. California is advancing SB 947, requiring human oversight of automated employment decisions [14][15], and SB 951, requiring 90-day advance notice of AI-driven layoffs [16][17]. Florida AG Uthmeier filed a civil complaint against OpenAI and Sam Altman arguing the company marketed ChatGPT as safe while knowing it posed harm to children [18], and separately opened a criminal investigation tied to the FSU mass shooting [19][20] — a litigation track that likely operates outside the preemption EO's reach, which was written to target state legislation rather than tort claims.

OpenAI is now an explicit advocate for federal AI governance, publishing a comprehensive public policy agenda [21] and a blueprint for democratic governance of frontier AI framing national security as the central justification for a coherent federal framework rather than fragmented state rules [22]. Semafor reports OpenAI is pursuing a state-level regulatory strategy intended to create a de facto national governance framework in the absence of federal legislation [23]. Senator Sanders' American AI Sovereign Wealth Fund Act — a proposed one-time 50% stock transfer from major AI companies to a publicly managed sovereign fund [24] — drew broad public attention after a NYT op-ed framing AI wealth as built on collective public knowledge [25], but has no reported co-sponsors and is unlikely to pass the current Congress.

Timeline

  • 2025-12: Trump signs executive order 'Ensuring a National Policy Framework for Artificial Intelligence,' aimed at eliminating state law obstruction of national AI policy [27][2][31]
  • 2026-01-01: New state AI laws take effect nationally, triggering the federal preemption debate [62]
  • 2026-03-30: California issues Executive Order N-5-26 establishing AI certification and procurement standards for state government agencies [37][38][39]
  • 2026-05: Trump signs executive order preempting or challenging state AI laws; multiple law firms find the EO's constitutional authority limited absent congressional authorization [1][2][3][4][5][49][50][51][52]
  • 2026-05: California SB 951 advances requiring 90-day advance notice of AI-driven layoffs, backed by California Federation of Labor [16][17][63]
  • 2026-05-21: Trump cancels AI security EO after Sacks calls Trump directly without staff knowledge to oppose pre-deployment review; Musk and Zuckerberg also lobby against it; OpenAI executives were mid-air to Washington when event is canceled [6][7][30][28][29]
  • 2026-05-21: California Governor Newsom signs AI workforce executive order directing state agencies to study severance pay and workforce support for AI-displaced workers [34][35][36][40]
  • 2026-05: California Senate approves No Robo Bosses Act (SB 947), requiring human oversight of automated decision systems in employer discipline and termination [14][15][59][60]
  • 2026-05: Representatives Obernolte (R-CA) and Jacobs (D-CA) introduce bipartisan AI Workforce PREPARE Act in 119th Congress [47][44][45]
  • 2026-05-28: OpenAI publishes Frontier Governance Framework aligned with EU and California requirements and a transparency statement supporting thoughtful AI regulation [32][33]
  • 2026-05-28: Illinois Governor Pritzker signs SB 315 — the nation's strongest AI safety statute, requiring mandatory third-party auditing, 72-hour critical incident reporting, and whistleblower protections [11][7][12][41][13]
  • 2026-06-01: Florida becomes first state to sue OpenAI and Sam Altman over ChatGPT's alleged role in violent crimes including the FSU mass shooting; AG Uthmeier also opens a separate criminal investigation [42][18][19][26][20][43]
  • 2026-06-02: Senator Sanders announces plans to introduce the American AI Sovereign Wealth Fund Act, proposing a one-time 50% stock transfer from major AI companies to a public sovereign fund [48][24][25]
  • 2026-06-03: Trump signs revised AI executive order establishing a voluntary 30-day pre-release review framework for frontier models, with no binding requirements; critics argue it is de facto mandatory and DOGE cuts limit evaluation capacity [8][9][10]
  • 2026-06-03: OpenAI publishes a public policy agenda and a blueprint for democratic governance of frontier AI, explicitly calling for a federal framework over fragmented state regulation and framing frontier AI as a national security matter [21][22][23]

Perspectives

Trump administration (federal)

Signed a revised AI EO establishing a voluntary 30-day pre-release review for frontier models, framed as balancing safety with innovation; the EO establishes no binding requirements and explicitly disclaims creating a licensing regime.

Evolution: Changed since the May cancellation: signed a significantly weakened version of the canceled order after accelerationist lobbying reduced the window from 90 to 30 days and stripped binding requirements.

[27][1][2][28][29][30][31][6][8][9][10]

Tech accelerationist faction (Musk, Zuckerberg, Sacks)

Actively opposes AI oversight mechanisms; successfully lobbied to cancel the original AI security EO and diluted the signed version to a voluntary 30-day window with no binding requirements.

Evolution: Consistent; Zvi Mowshowitz notes this faction won the policy fight on structure even as the EO was eventually signed.

[6][7][8]

OpenAI

Explicitly supports a coherent federal governance framework; published a public policy agenda and a blueprint framing frontier AI as a national security matter requiring federal — not fragmented state — oversight.

Evolution: Stance has evolved: moved from publishing safety documents to explicitly calling for federal governance architecture to displace state-level fragmentation, sharpening the split from the accelerationist faction.

[6][32][33][21][22][23]

State governors advancing AI legislation (Newsom + Pritzker)

State government must proactively address AI harms through legislation; Illinois SB 315 is the nation's strictest enacted AI safety statute; California is advancing SB 947 and SB 951 with workforce executive orders and procurement standards.

Evolution: Consistent; the two-state legislative coalition continues operating independently of federal coordination.

[34][35][36][37][38][39][16][17][14][15][40][11][12][41][13]

Florida AG Uthmeier (litigation track)

Pursues AI accountability through civil tort law and criminal investigation; argues OpenAI marketed ChatGPT as safe while knowing it posed harm to children — a 'knew of harm' theory targeting product liability rather than general negligence.

Evolution: Consistent; the litigation track remains structurally distinct from the legislative track and likely operates outside the preemption EO's reach.

[42][18][19][26][20][43]

Congress — federal legislative responses

Obernolte and Jacobs introduced the bipartisan PREPARE Act framing AI workforce concerns as development challenges; Sanders plans a more radical American AI Sovereign Wealth Fund Act proposing a 50% stock transfer for public ownership — a wide spectrum with neither likely to pass the current Congress.

Evolution: Consistent; Sanders' proposal has generated extensive public attention but no reported co-sponsors.

[44][45][46][47][48][24]

Analysts critical of the Trump AI EO (Zvi Mowshowitz, Ars Technica)

The EO is either de facto mandatory despite its 'voluntary' framing with classified thresholds that prevent labs from knowing their own compliance status (Zvi), or too weak to prevent dangerous deployments given DOGE-gutted evaluation capacity (Ars Technica).

Evolution: New voice this pass, responding directly to the signed EO.

[8][10]

Legal and employment law analysts

The preemption EO's constitutional authority is ambiguous absent congressional authorization; California's SB 947 and SB 951 create growing employer compliance obligations reaching automated decision systems in individual employment decisions.

Evolution: Consistent.

[3][4][5][49][50][51][52][53][54][55][56][57][14][15][58][59][60][61]

Tensions

  • Voluntary vs. de facto mandatory: The Trump AI EO disclaims creating a licensing regime, but Zvi Mowshowitz argues labs have no practical choice but to participate — and that classifying benchmarking thresholds prevents labs from knowing if their own models are subject to review [8]. [8][9]
  • Stated mandate vs. enforcement capacity: Trump's EO establishes a review process for frontier models, but DOGE-driven cuts to federal cybersecurity and safety teams have reduced the government's practical capacity to conduct real evaluations [10]. [10][8]
  • Accelerationist faction vs. pro-regulation OpenAI: Musk, Zuckerberg, and Sacks successfully diluted the signed EO to a voluntary 30-day window, while OpenAI explicitly supports a federal governance framework and national security framing for frontier AI [6][7][21]. [6][7][21][22]
  • Federal preemption vs. state governance: The Trump preemption EO claims to block state AI laws, but multiple law firms find its constitutional authority limited absent congressional authorization, potentially leaving state laws intact until courts rule [1][3]. [1][2][3][5][49]
  • State legislative mandates vs. state litigation: California and Illinois advance statutory AI mandates while Florida pursues OpenAI through tort and criminal tracks that likely bypass both legislative processes and the preemption EO [11][42][19]. [11][42][19]
  • OpenAI's federal framework preference vs. the current state-driven landscape: OpenAI advocates for a coherent federal governance architecture that would displace fragmented state rules [22][23], while binding state laws from California and Illinois are advancing without federal coordination [11][13]. [21][22][23][11][13]

Sources

  1. [1] President Trump Signs Executive Order Preempting State AI Laws ... — reactive:us-ai-policy-regulation
  2. [2] President Trump Signs Executive Order to Block State AI Laws — reactive:us-ai-policy-regulation
  3. [3] President Trump Signs Executive Order Challenging State AI Laws | Paul Hastings LLP — reactive:us-ai-policy-regulation
  4. [4] When Federal Preemption Meets AI Regulation: What Trump's Draft Executive Order Means for Your Compliance Strategy | Jones Walker LLP — reactive:us-ai-policy-regulation
  5. [5] Executive Preemption and the Dormant Commerce Clause After ... — reactive:us-ai-policy-regulation
  6. [6] Trump abruptly cancels EO signing event after top AI firm CEOs declined to go — Ars Technica AI (2026-05-22)
  7. [7] AI #170: Lack of Executive Order — Zvi's AI Roundups (2026-05-28)
  8. [8] Trump Signs Executive Order For AI Testing Prior To Frontier Model Releases — Zvi's AI Roundups (2026-06-03)
  9. [9] President Trump signs executive order to review advanced AI models. — Rohan Paul Twitter (2026-06-03)
  10. [10] Trump plan to test AI models has a problem—US security teams were gutted by DOGE — Ars Technica AI (2026-06-03)
  11. [11] Trump loses more control over AI regulation as Illinois passes landmark law — Ars Technica AI (2026-05-28)
  12. [12] Illinois Legislature passes historic AI bill that would require third-party safety audits — reactive:us-ai-policy-regulation
  13. [13] Illinois Signs Frontier AI Safety Bill into Law | Chris Lehane posted ... — reactive:us-ai-policy-regulation
  14. [14] CA Senate Approves No Robo Bosses Act of 2026 to Ensure Human Oversight of AI in the Workplace | Senator Jerry McNerney — reactive:us-ai-policy-regulation
  15. [15] California Set to Restrict AI Use in the Workplace With “No Robo ... — reactive:us-ai-policy-regulation
  16. [16] Bill Text: CA SB951 | 2025-2026 | Regular Session | Amended — reactive:us-ai-policy-regulation
  17. [17] SB 951: Employment: technological displacement: notice. — reactive:us-ai-policy-regulation
  18. [18] ChatGPT creators knew product would cause harm, Florida argues in lawsuit • Florida Phoenix — reactive:us-ai-policy-regulation
  19. [19] Attorney General James Uthmeier Launches Criminal Investigation into OpenAI, ChatGPT | My Florida Legal — reactive:us-ai-policy-regulation
  20. [20] OpenAI let ChatGPT aid and abet mass shooters, Florida suit says — reactive:us-ai-policy-regulation
  21. [21] OpenAI public policy agenda — OpenAI Blog (2026-06-03)
  22. [22] A blueprint for democratic governance of frontier AI — OpenAI Blog (2026-06-03)
  23. [23] 🟡 Let’s get physical — Semafor Technology (2026-06-03)
  24. [24] Sanders Sovereign Wealth Fund Plan Would Give US Public 'Direct Ownership Stake' in AI Giants | Common Dreams — reactive:us-ai-policy-regulation
  25. [25] Sen. Bernie Sanders on X: "AI is built on humanity’s collective knowledge. The wealth it generates must benefit humanity — not just Elon Musk, Sam Altman and other AI oligarchs. That’s why I’ll be introducing the American AI Sovereign Wealth Fund Act — to give the public a direct ownership stake. https://t.co/UqW71FBv2Z" / X — reactive:us-ai-policy-regulation
  26. [26] Florida sues OpenAI and CEO Sam Altman, accusing them of putting profit over safety - NBC News — reactive:us-ai-policy-regulation
  27. [27] Ensuring a National Policy Framework for Artificial Intelligence — reactive:us-ai-policy-regulation
  28. [28] Trump delays executive order on AI oversight hours before planned ... — reactive:us-ai-policy-regulation
  29. [29] Trump scraps signing of landmark executive order regulating AI — reactive:us-ai-policy-regulation
  30. [30] WATCH: Trump explains why he postponed signing AI executive order — reactive:us-ai-policy-regulation
  31. [31] Executive Order: Ensuring a National Policy Framework for Artificial Intelligence (Donald Trump, 2025) - Ballotpedia — reactive:us-ai-policy-regulation
  32. [32] OpenAI’s Frontier Governance Framework — OpenAI Blog (2026-05-28)
  33. [33] Our views on AI policy and political advocacy — OpenAI Blog (2026-06-01)
  34. [34] [PDF] EXECUTIVE DEPARTMENT STATE OF CALIFORNIA — reactive:ai-labor-market-debate
  35. [35] California governor orders official to find ways to mitigate AI layoffs — reactive:ai-labor-market-debate
  36. [36] Gov. Newsom signs executive order directing agencies to prepare for AI job disruptions. UC Davis professor reacts — reactive:ai-labor-market-debate
  37. [37] [PDF] executive order (N-5-26) - Governor of California — reactive:us-ai-policy-regulation
  38. [38] Newsom Signs Executive Order Establishing AI Vendor Certification ... — reactive:us-ai-policy-regulation
  39. [39] California Executive Order N-5-26 — Responsible Procurement and ... — reactive:us-ai-policy-regulation
  40. [40] California eyes AI regulation as Gov. Newsom orders new workforce protections amid job shifts, mass layoffs - ABC7 San Francisco — reactive:us-ai-policy-regulation
  41. [41] Illinois Lawmakers Just Passed America's Strongest AI Safety Bill — reactive:us-ai-policy-regulation
  42. [42] Florida sues OpenAI, Sam Altman after multiple ChatGPT-linked murders — Ars Technica AI (2026-06-01)
  43. [43] Florida sues OpenAI and Sam Altman over AI risks — reactive:us-ai-policy-regulation
  44. [44] Text - S.3339 - 119th Congress (2025-2026): AI Workforce PREPARE Act — reactive:us-ai-policy-regulation
  45. [45] Rep. Obernolte, Rep. Jacobs Introduce Bipartisan Bill to Prepare American Workers for AI-Driven Economic Change | Representative Jay Obernolte — reactive:us-ai-policy-regulation
  46. [46] S3339 | US Congress 2025-2026 | AI Workforce PREPARE Act — reactive:us-ai-policy-regulation
  47. [47] S.3339 - 119th Congress (2025-2026): AI Workforce PREPARE Act — reactive:us-ai-policy-regulation
  48. [48] Bernie Sanders announces he will soon introduce the American AI Sovereign Wealth Fund Act. — Rohan Paul Twitter (2026-06-02)
  49. [49] Legal Issues Raised by the Proposed Executive Order on AI Preemption - Institute for Law & AI — reactive:us-ai-policy-regulation
  50. [50] AI Executive Order: Litigation & Preemption FAQ — reactive:us-ai-policy-regulation
  51. [51] Decoding the 2026 White House AI Blueprint: U.S. AI Policy Starts to ... — reactive:us-ai-policy-regulation
  52. [52] California’s New Executive Order Establishes New AI Vendor Certification and Procurement Requirements | Vinson & Elkins LLP - JDSupra — reactive:us-ai-policy-regulation
  53. [53] State Laws Impacting Employers in 2026: Amundsen Davis — reactive:us-ai-policy-regulation
  54. [54] Employment Laws on the Horizon Report | Seyfarth Shaw LLP — reactive:us-ai-policy-regulation
  55. [55] A Concerted Effort to Regulate Workplace Technology – What Public Employers Need to Know About Proposed State Legislation - Liebert Cassidy Whitmore — reactive:us-ai-policy-regulation
  56. [56] California Enacts Strict AI Employment Rules: SB 947 & SB 951 | Lauren Goetzl posted on the topic | LinkedIn — reactive:us-ai-policy-regulation
  57. [57] California Employment Law Update for 2026 | HUB | K&L Gates — reactive:us-ai-policy-regulation
  58. [58] The California “No Robo Bosses Act” - American Society of Employers — reactive:us-ai-policy-regulation
  59. [59] California SB 947 ("No Robo Bosses Act"): New Proposed Guardrails on Use of Automated Decision Systems in Employer Discipline and Termination Decisions | Crowell & Moring LLP — reactive:us-ai-policy-regulation
  60. [60] California Set to Restrict AI Use in the Workplace With “No Robo Bosses” Act: 4 Key Steps Employers Should Take to Comply | Fisher Phillips - JDSupra — reactive:us-ai-policy-regulation
  61. [61] Trump Moves to Pre-empt State AI Laws with Executive Order - SHRM — reactive:us-ai-policy-regulation
  62. [62] New State AI Laws are Effective on January 1, 2026, But a New ... — reactive:us-ai-policy-regulation
  63. [63] California Labor Unions Demand Transparency and Human Oversight of Artificial Intelligence with New Legislation - California Federation of Labor Unions — reactive:us-ai-policy-regulation