The Information Machine

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

Version 27

2026-06-28 08:21 UTC · 376 items

What

The Trump administration's request that OpenAI limit GPT-5.6 to government-approved customers has moved from request to implementation: OpenAI released GPT-5.6 as a controlled preview with the government acting as access gatekeeper [12][13], backed by a system card published under OpenAI's Deployment Safety Hub [29][30]. The administration continues pressing Meta — the only major US lab outside its voluntary review system — toward participation, with the White House stating 'We hope to sign the agreement soon' [9]. Critics argue that restricting US frontier model access through government gatekeeping may push price-sensitive developers toward Chinese and open-weight alternatives [14][15], a dynamic that cuts against the administration's stated national security rationale.

Why it matters

The government-gated GPT-5.6 launch is the first concrete instance of executive-branch control over post-deployment frontier AI access, achieved without statutory authority and beyond the terms of the June 3 executive order. If developers respond by migrating to ungated alternatives, the access control strategy could produce outcomes contrary to its stated purpose of maintaining US AI advantage.

Open questions

  • Will the government-gated GPT-5.6 rollout push developers toward Chinese or open-weight models, undermining the US compute-dominance rationale? [14][15]

  • Does the customer-by-customer approval role the government took for GPT-5.6 extend to future frontier models, or is it specific to models with demonstrated autonomous cyberattack capabilities? [11][12]

  • Will Meta comply with the White House's confidence that an agreement is imminent [9], and will the terms of Meta's participation differ from what OpenAI accepted?

  • Will the 42-state AG coalition subpoena [25] and Florida's personal suit against Altman [26] converge on coordinated settlement terms or remain parallel enforcement tracks?

Narrative

The Trump administration has constructed an executive-action oversight regime for frontier AI without legislation. Its June 3, 2026 executive order established a voluntary 30-day pre-release review for frontier models with NSA oversight and classified compliance thresholds, explicitly disclaiming binding requirements [1][2][3]. That narrow form reflects direct lobbying by Elon Musk, Mark Zuckerberg, and White House AI czar David Sacks, who opposed binding requirements and shortened the review window from 90 to 30 days [4][5]. Since signing, the administration has applied the framework coercively: it forced Anthropic to disable Fable-class models for all foreign nationals [6][7], moved the DOJ to protect xAI's Mississippi data center as national security infrastructure [8], and pressed Meta — the only major US lab outside the review system — to join, with the White House stating 'We hope to sign the agreement soon' [9]. One observer noted the administration has assembled the institutional pieces of an AI oversight apparatus — classified capability thresholds, pre-release government access, and trusted-lab designations — through executive action alone, without the statutory foundation that would normally authorize such control [10].

The most direct expression of this posture is the government-gated launch of GPT-5.6. The administration asked OpenAI to release the model as a controlled, staggered preview with customer-by-customer government approval, citing cybersecurity concerns about the model's autonomous capabilities for offensive cyber operations [11]. OpenAI complied: GPT-5.6 is now available only through a controlled preview with the government serving as access gatekeeper [12][13]. This is the first concrete instance of executive-branch control over post-deployment frontier AI access — a step the June 3 executive order did not authorize and for which no statutory basis exists. Critics note a strategic tension: restricting access to leading US frontier models may push price- and latency-sensitive developers toward Chinese or open-weight alternatives [14][15], running counter to the administration's stated goal of maintaining US AI dominance.

The public argument over AI oversight has centered on an exchange between Anthropic CEO Dario Amodei and the administration's accelerationist wing. Amodei called for mandatory pre-release testing and government-backed independent auditing [16][17]. Sacks accused Anthropic of 'a sophisticated regulatory capture strategy based on fear-mongering' [18]; VP Vance echoed the charge [19]. The administration simultaneously made Anthropic the subject of the Fable export control action, and Bloomberg reported that Anthropic's AI principles have made it a White House target [20]. Former Google CEO Eric Schmidt supplied the public rationale: the US controls approximately 75% of global AI compute versus China's 15%, a gap he attributes to coordinated export controls [21].

State governments advance binding AI regulation without federal coordination. Illinois enacted SB 315 requiring annual third-party audits, 72-hour incident reporting, and whistleblower protections [22]. California is advancing SB 947 and SB 951 requiring human oversight of automated employer discipline and 90-day advance notice of AI-driven layoffs [23][24]. New York AG Letitia James, on behalf of a 42-state coalition, subpoenaed OpenAI over consumer practices, data handling, treatment of minors, and chatbot sycophancy [25]; Florida expanded its lawsuit to 83 pages naming CEO Sam Altman personally [26]. Trump's preemption executive order challenged state AI laws on constitutional grounds [27], but multiple law firms found its authority limited absent congressional action [28].

Timeline

  • 2026-05: Trump signs preemption EO challenging state AI laws; multiple law firms find constitutional authority limited absent congressional action [27][65][28]
  • 2026-05: California Senate advances SB 947 requiring human oversight of automated employer discipline and SB 951 requiring 90-day advance notice of AI-driven layoffs [24][47][23][48]
  • 2026-05-21: Trump cancels AI security EO after Sacks calls Trump directly to oppose pre-deployment review; Musk and Zuckerberg also lobby against binding requirements [4][5][66]
  • 2026-05-28: Illinois Governor Pritzker signs SB 315 requiring mandatory third-party auditing, 72-hour critical incident reporting, and whistleblower protections [49][50][22]
  • 2026-06-01: Florida becomes first state to sue OpenAI and Sam Altman over ChatGPT's alleged role in violent crimes; AG Uthmeier opens a separate criminal investigation [51][52][53]
  • 2026-06-03: Trump signs revised AI executive order establishing a voluntary 30-day pre-release review for frontier models with NSA oversight and classified compliance thresholds [1][2][31][3]
  • 2026-06-04: Build American AI — backed by OpenAI, a16z, and Palantir — acknowledged running false-flag social media accounts; WIRED reports the PAC paid TikTok influencers to fear-monger about Chinese AI [44][45][67][68]
  • 2026-06-10: Anthropic CEO Dario Amodei publishes blog calling for mandatory pre-release testing and government-backed independent auditing [16][17]
  • 2026-06-12: New York AG Letitia James serves OpenAI with a subpoena on behalf of a 42-state coalition, targeting advertising practices, data handling, treatment of minors, and chatbot sycophancy [25]
  • 2026-06-14: David Sacks publicly accuses Anthropic of 'regulatory capture strategy based on fear-mongering'; Amodei responds defending his regulatory stance as aligned with Trump priorities [34][18][36][37][38]
  • 2026-06-14: US government forces Anthropic to disable its Fable-class AI models for all foreign nationals [6][39][7]
  • 2026-06-15: Florida expands its lawsuit against OpenAI to 83 pages naming CEO Sam Altman personally, separate from the ongoing criminal probe [26][54]
  • 2026-06-18: DOJ moves to protect xAI's Mississippi AI data center, treating high-speed AI compute as national security infrastructure [8]
  • 2026-06-19: VP Vance publicly accuses AI CEOs of using fear as a marketing strategy, echoing Sacks' critique of AI safety advocacy [19]
  • 2026-06-21: Bloomberg reports Anthropic's AI safety principles have made it a White House target; Eric Schmidt states the US controls ~75% of global AI compute vs China's 15%, attributing the gap to coordinated export controls [20][21]
  • 2026-06-24: Trump administration pressures Meta to submit AI models to government review before public release; Meta is the only major US AI lab outside the voluntary review system [32][58][33][60]
  • 2026-06-25: US government asks OpenAI to release GPT-5.6 as a controlled preview with customer-by-customer government approval, citing cybersecurity concerns; White House says 'We hope to sign the agreement soon' about Meta joining the review system [11][9]
  • 2026-06-27: GPT-5.6 launches as a government-gated controlled preview; OpenAI publishes a system card under its Deployment Safety Hub [12][13][29][30]

Perspectives

Trump administration (federal)

Signed a voluntary AI EO with NSA oversight disclaiming binding requirements; simultaneously forced Anthropic to disable Fable-class models for foreign nationals, moved DOJ to protect xAI infrastructure as national security property, pressed Meta to join the review system, and secured a government-gated launch for GPT-5.6.

Evolution: The administration has moved from pre-release review to post-deployment access control: the GPT-5.6 government-gated launch extends executive oversight beyond what the June 3 executive order authorized and without statutory basis.

[27][4][1][2][31][3][8][6][20][32][33][11][9][12]

Tech accelerationist faction (Sacks, Musk, Zuckerberg, Vance)

Opposes domestic AI oversight mechanisms; Sacks accuses Anthropic of regulatory capture through fear-mongering; VP Vance echoed the framing publicly.

Evolution: Consistent in opposing domestic binding requirements; Sacks reportedly seeking to narrow his critique to specific security concerns around Mythos-class models rather than general AI safety advocacy.

[4][5][34][18][19][35]

Anthropic / Dario Amodei

Calls for mandatory pre-release testing and government-backed independent auditing; disputes the 'fear-mongering' characterization while claiming alignment with Trump administration priorities.

Evolution: Confirmed subject of a government export control action requiring it to disable Fable-class models for foreign nationals; Bloomberg characterizes its AI principles as making it a White House target — a difficult position as Amodei simultaneously defends his regulatory advocacy as pro-administration.

[17][16][36][37][38][6][39][20][7]

OpenAI

Publicly advocates for federal governance and shared evaluation standards; complied with the government's request to launch GPT-5.6 as a controlled, government-gated preview.

Evolution: The GPT-5.6 government-gated launch represents compliance with executive access control that goes beyond pre-release review — testing whether OpenAI's stated support for federal governance extends to accepting access management with no clear statutory basis.

[40][41][42][43][44][45][46][25][11][12][13][29]

State legislators (Illinois + California)

Advancing binding AI safety and employment law independent of federal coordination; Illinois SB 315 enacted; California SB 947 and SB 951 advancing.

Evolution: Consistent; operating without federal coordination and not slowed by the Trump preemption EO.

[24][47][23][48][49][50][22]

State AGs (42-state coalition + Florida litigation)

Pursues AI accountability through civil and criminal enforcement; the 42-state coalition subpoenaed OpenAI over consumer practices; Florida expanded its suit to 83 pages naming Altman personally.

Evolution: Expanded from Florida's single-state lawsuit to a coordinated 42-state investigation; legal theorists including Bruce Schneier have publicly argued for deployer liability frameworks consistent with the AGs' enforcement theory.

[51][52][53][25][26][54][55][56][57]

Meta

Has not joined the voluntary pre-release review system; the administration is pressing it to participate, with the White House expressing public confidence that an agreement is imminent.

Evolution: The White House's public confidence framing suggests the administration expects compliance rather than confrontation — Meta's actual response remains publicly unconfirmed.

[32][58][59][33][60][61][62][63][9][64]

Eric Schmidt

Publicly confirmed that US export controls have produced a 75%-to-15% US-China compute gap, providing the clearest public articulation of the national security doctrine behind administration AI policy.

Evolution: Consistent; provides the most direct public rationale for the enforcement logic the administration has acted on but not stated explicitly.

[21]

Tensions

  • Mandatory oversight (Amodei) vs. accelerationist opposition (Sacks, Vance): Amodei calls for mandatory pre-release testing and independent auditing [16]; Sacks and VP Vance argue AI safety advocates manufacture fear for commercial reasons [18][19] — yet the administration simultaneously disabled Anthropic's Fable models for foreign nationals and Bloomberg reports Anthropic's principles make it a White House target [6][20]. [16][18][6][19][20]
  • Federal preemption preference vs. state enforcement reality: Trump's preemption EO lacks constitutional authority absent congressional action [27][28]; binding Illinois law [22] and a 42-state coalition subpoena [25] advance without federal coordination. [27][28][22][25]
  • National security access control vs. risk of accelerating competitor adoption: the government has gated GPT-5.6 behind customer-by-customer approval citing cybersecurity [11][12], but critics argue this may push developers toward Chinese and open-weight alternatives [14][15], working against the US compute-dominance rationale Schmidt articulated [21]. [11][12][14][15][21]
  • 'Voluntary' review framework vs. coercive application: the administration labeled the review system voluntary [1][3] but is pressing Meta to join [9] and secured a government-gated launch for GPT-5.6 [12][13] — neither action is contemplated in the voluntary framework's stated terms. [1][3][9][12][13]
  • OpenAI's governance advocacy vs. its PAC's political operations and legal exposure: OpenAI calls for democratic governance and shared evaluation standards [41] while Build American AI admitted running sockpuppet accounts [46] and the 42-state subpoena targets the consumer practices OpenAI presents as responsible [25]. [41][44][46][25]
  • Government access control vs. absence of statutory authority: the administration's role as access gatekeeper for GPT-5.6 [12][13] extends executive oversight into post-deployment access management without any clear statutory basis, going beyond what the June 3 EO authorized [1][3]. [12][13][1][3]

Sources

  1. [1] Trump Signs Executive Order For AI Testing Prior To Frontier Model Releases — Zvi's AI Roundups (2026-06-03)
  2. [2] President Trump signs executive order to review advanced AI models. — Rohan Paul Twitter (2026-06-03)
  3. [3] Executive Order Creates Voluntary Regulatory Regime of Frontier AI ... — reactive:us-ai-policy-regulation
  4. [4] Trump abruptly cancels EO signing event after top AI firm CEOs declined to go — Ars Technica AI (2026-05-22)
  5. [5] AI #170: Lack of Executive Order — Zvi's AI Roundups (2026-05-28)
  6. [6] The US government just forced Anthropic to disable its most advanced AI models for all foreign nationals — and the compa... — reactive:us-ai-policy-regulation (2026-06-14)
  7. [7] David Sacks on the US government's shutdown of Anthropic's Fable ... — reactive:us-ai-policy-regulation
  8. [8] Large AI campuses are indeed strategic infrastructure of a nation. — Rohan Paul Twitter (2026-06-18)
  9. [9] 'We hope to sign the agreement soon': White House calls on Meta to submit AI models for review, citing abilities and vulnerabilities evaluation | TechRadar — reactive:us-ai-policy-regulation
  10. [10] it has created the institutional pieces of one: classified capability thresholds, pre-release government access, “truste... — reactive:g7-ai-frontier-summit (2026-06-26)
  11. [11] The Information: The US government is asking OpenAI to slow GPT-5.6 into a controlled preview instead of releasing it br… — Rohan Paul Twitter (2026-06-25)
  12. [12] OpenAI Limits GPT-5.6 Rollout at US Government's Request — reactive:us-ai-policy-regulation
  13. [13] OpenAI releases GPT-5.6 with government as gatekeeper - LinkedIn — reactive:us-ai-policy-regulation
  14. [14] The shift toward Chinese/open-weight models was already happening because developers follow price, latency, availability... — reactive:gpt-56-launch-government-access (2026-06-26)
  15. [15] China’s open-weight strategy is no longer just “catch-up.” It is becoming a release-cadence weapon. This month, Chinese ... — reactive:us-ai-policy-regulation (2026-06-26)
  16. [16] Dario Amodei just published a super long blog, calling for an urgent policy overhaul because he thinks frontier AI is mo… — Rohan Paul Twitter (2026-06-10)
  17. [17] Dario Amodei argues in his new blog today, that powerful AI will soon become the dominant source of military and economi… — Rohan Paul Twitter (2026-06-11)
  18. [18] Anthropic is running a sophisticated regulatory capture strategy based on fear-mongering. It is principally responsible for the state regulatory frenzy that is damaging the startup ecosystem. — reactive:us-ai-policy-regulation
  19. [19] US VP JD Vance on AI Company's marketing strategy, some AI CEOs sell fear because fear makes their product look powerful… — Rohan Paul Twitter (2026-06-19)
  20. [20] Bloomberg, “Anthropic's AI Principles Make It a White House Target” — reactive:us-ai-policy-regulation
  21. [21] Eric Schmidt just said in public what US AI policy has been built around in private for three years (Save this). — Milk Road AI Twitter (2026-06-21)
  22. [22] Illinois Signs Frontier AI Safety Bill into Law | Chris Lehane posted ... — reactive:us-ai-policy-regulation
  23. [23] 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
  24. [24] Bill Text: CA SB951 | 2025-2026 | Regular Session | Amended — reactive:us-ai-policy-regulation
  25. [25] 😺 42 states just subpoenaed OpenAI — The Neuron (2026-06-15)
  26. [26] Florida AG Sues OpenAI, Alleges ChatGPT Aided Mass Shooters — reactive:us-ai-policy-regulation
  27. [27] President Trump Signs Executive Order Preempting State AI Laws ... — reactive:us-ai-policy-regulation
  28. [28] President Trump Signs Executive Order Challenging State AI Laws | Paul Hastings LLP — reactive:us-ai-policy-regulation
  29. [29] GPT-5.6 Preview System Card - OpenAI Deployment Safety Hub — reactive:us-ai-policy-regulation
  30. [30] [PDF] GPT-5.6 Preview System Card - Deployment Safety Hub — reactive:us-ai-policy-regulation
  31. [31] Trump AI Executive Order Gives NSA Classified Role Over Frontier Models — reactive:us-ai-policy-regulation
  32. [32] NYT: The Trump administration is pressuring Meta to submit its AI models to government review before public release. — Rohan Paul Twitter (2026-06-24)
  33. [33] Meta Faces Growing Pressure From Trump Administration To Submit Advanced AI Models For Federal Testing: R - Benzinga — reactive:us-ai-policy-regulation
  34. [34] David Sacks is done being polite about Anthropic (Save this). — Milk Road AI Twitter (2026-06-14)
  35. [35] Sacks urges focus on Mythos-driven security needs, seeks to ‘correct’ the narrative around his Anthropic critique | InsideAIPolicy.com — reactive:us-ai-policy-regulation
  36. [36] Anthropic CEO disputes David Sacks' claims that company is 'woke' — reactive:us-ai-policy-regulation
  37. [37] Anthropic CEO defends support for AI regulations, alignment with ... — reactive:us-ai-policy-regulation
  38. [38] Anthropic CEO claps back after Trump officials accuse firm of AI fear-mongering   | TechCrunch — reactive:us-ai-policy-regulation
  39. [39] Legal Considerations Related to the Anthropic “Export Controls ... — reactive:us-ai-policy-regulation
  40. [40] OpenAI’s Frontier Governance Framework — OpenAI Blog (2026-05-28)
  41. [41] OpenAI public policy agenda — OpenAI Blog (2026-06-03)
  42. [42] A blueprint for democratic governance of frontier AI — OpenAI Blog (2026-06-03)
  43. [43] 🟡 Let’s get physical — Semafor Technology (2026-06-03)
  44. [44] AI #171: False Flag — Zvi's AI Roundups (2026-06-04)
  45. [45] The OpenAI-Andreessen-Palantir SuperPAC admits that it was “part ... — reactive:us-ai-policy-regulation
  46. [46] Build American AI admits hiring an outside vendor to run anonymous sockpuppet accounts targeting its opponents · Digg — reactive:us-ai-policy-regulation
  47. [47] SB 951: Employment: technological displacement: notice. — reactive:us-ai-policy-regulation
  48. [48] California Set to Restrict AI Use in the Workplace With “No Robo ... — reactive:us-ai-policy-regulation
  49. [49] Trump loses more control over AI regulation as Illinois passes landmark law — Ars Technica AI (2026-05-28)
  50. [50] Illinois Legislature passes historic AI bill that would require third-party safety audits — reactive:us-ai-policy-regulation
  51. [51] ChatGPT creators knew product would cause harm, Florida argues in lawsuit • Florida Phoenix — reactive:us-ai-policy-regulation
  52. [52] Attorney General James Uthmeier Launches Criminal Investigation into OpenAI, ChatGPT | My Florida Legal — reactive:us-ai-policy-regulation
  53. [53] Florida AG brings 'first-in-the-nation state-led lawsuit' against OpenAI and its CEO — reactive:us-ai-policy-regulation
  54. [54] Florida Attorney General Sues OpenAI Over ChatGPT's Alleged ... — reactive:us-ai-policy-regulation
  55. [55] Florida attorney general sues OpenAI | Courthouse News Service — reactive:us-ai-policy-regulation
  56. [56] Florida sues OpenAI over ChatGPT risks Attorney General ... — reactive:us-ai-policy-regulation
  57. [57] AI and Liability — Simon Willison (2026-06-25)
  58. [58] Meta Faces White House Push to Join AI Pre-Release ... — reactive:us-ai-policy-regulation
  59. [59] US Government Pressures Meta to Allow Pre-Release Reviews of Its ... — reactive:us-ai-policy-regulation
  60. [60] Trump administration pushes Meta for AI model security reviews- NYT — reactive:us-ai-policy-regulation
  61. [61] The Trump administration is leaning on Meta to agree to voluntary government review of its AI models — and as of today, ... — reactive:us-ai-policy-regulation (2026-06-24)
  62. [62] Meta Under Pressure to Join Federal AI Model Review Program - Techstrong.ai — reactive:us-ai-policy-regulation
  63. [63] Trump Administration Presses Meta To Submit AI Models For Federal Review — reactive:us-ai-policy-regulation
  64. [64] US Government Reportedly Urging Meta To Share Its AI Models — reactive:us-ai-policy-regulation
  65. [65] President Trump Signs Executive Order to Block State AI Laws — reactive:us-ai-policy-regulation
  66. [66] WATCH: Trump explains why he postponed signing AI executive order — reactive:us-ai-policy-regulation
  67. [67] A Pro-AI Super PAC’s Secret Meme Sockpuppets — reactive:us-ai-policy-regulation
  68. [68] A Dark-Money Campaign Is Paying Influencers to Frame Chinese AI as a Threat | WIRED — reactive:us-ai-policy-regulation