The Information Machine

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

Version 29

2026-07-02 02:24 UTC · 409 items

What

The Trump administration has assembled AI oversight through executive action: a June 3 voluntary pre-release review with NSA oversight [1][3] and a government-gated launch of GPT-5.6 Sol in which the executive branch controls post-deployment customer access without statutory basis [15][16][17]. On June 30, the White House lifted the export control on Anthropic that had blocked its Fable-class models for foreign nationals since June 14 [8], without public explanation. Meta remains the only major US AI lab outside the voluntary review system [11][12][13]. State-level enforcement advances independently: Illinois enacted mandatory auditing [27], a 42-state AG coalition subpoenaed OpenAI [30], and Florida's suit names CEO Sam Altman personally [31].

Why it matters

The Anthropic export control — imposed and reversed within two weeks without public explanation — raises the question of whether the administration has stable enforcement tools or is improvising with leverage it cannot sustain. Meta's continued holdout remains the clearest test of whether the 'voluntary' review framework can be made universal through pressure alone, and whether the administration has any mechanism to compel compliance absent statutory authority.

Open questions

  • What prompted the White House to lift the Anthropic export control on June 30 [8] — did Anthropic make concessions, or was the ban temporary leverage that served its purpose?

  • Will the government's role as access gatekeeper for GPT-5.6 Sol [16][17] extend to future frontier releases, or is it limited to models with demonstrated autonomous offensive cyber capabilities?

  • Will Meta ultimately comply with the administration's review pressure [11][12][13], and if it refuses outright, does the administration have any enforcement mechanism beyond public statements?

  • Will the 42-state AG coalition's subpoena [30] and Florida's personal suit against Altman [31] converge on coordinated settlement terms or remain separate enforcement tracks?

Narrative

The Trump administration has built AI oversight through executive action, not legislation. Its June 3, 2026 executive order established a voluntary 30-day pre-release review for frontier AI 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 days to 30 [4][5].

The administration's practice has gone beyond its stated voluntary terms. It forced Anthropic to disable Fable-class models for all foreign nationals on June 14 [6][7], then reversed that action on June 30, lifting the export control without public explanation [8]. It moved the DOJ to protect xAI's Mississippi data center as national security infrastructure [9] and pressed Meta — the only major US AI lab still outside the review system — to join, while the White House publicly expressed confidence an agreement was imminent [10]; as of early July, Meta has not complied [11][12][13]. The clearest extension of executive control came with the launch of GPT-5.6 Sol: at the administration's request, OpenAI released the model as a controlled, government-gated preview with customer-by-customer approval, citing the model's autonomous offensive cyber capabilities [14][15][16][17]. Critics call this arrangement government-imposed 'AI feudalism' [18] and warn that access restrictions may push developers toward Chinese or open-weight alternatives [19][20], working against the US compute-dominance rationale Eric Schmidt articulated — approximately 75% US vs. 15% Chinese global AI compute, a gap Schmidt attributes to coordinated export controls [21].

The public argument over oversight has centered on a dispute between Anthropic CEO Dario Amodei and the administration's accelerationist faction. Amodei called for mandatory pre-release testing and government-backed independent auditing [22][23]; White House AI czar David Sacks accused Anthropic of 'a sophisticated regulatory capture strategy based on fear-mongering' [24], and VP Vance echoed that framing publicly [25]. Bloomberg reported that Anthropic's AI principles have made it a White House target [26] — a characterization complicated by the administration imposing and then lifting export controls on Anthropic's most capable models within two weeks [6][8].

State governments are advancing binding AI regulation without federal coordination. Illinois enacted SB 315 requiring annual third-party audits, 72-hour incident reporting, and whistleblower protections [27]. California is advancing SB 947 and SB 951 requiring human oversight of automated employer discipline and advance notice of AI-driven layoffs [28][29]. 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 [30]; Florida expanded its lawsuit to 83 pages naming CEO Sam Altman personally [31]. Trump's preemption executive order challenged state AI laws on constitutional grounds [32], but multiple law firms found its authority limited absent congressional action [33].

Timeline

  • 2026-05: Trump signs preemption EO challenging state AI laws; multiple law firms find constitutional authority limited absent congressional action [32][72][33]
  • 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 [29][60][28][61]
  • 2026-05-21: Trump cancels AI security EO after Sacks, Musk, and Zuckerberg lobby against binding pre-deployment review requirements [4][5][73]
  • 2026-05-28: Illinois Governor Pritzker signs SB 315 requiring mandatory third-party auditing, 72-hour critical incident reporting, and whistleblower protections [62][63][27]
  • 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 [64][65][66]
  • 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][34][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 [48][49][74][75]
  • 2026-06-10: Anthropic CEO Dario Amodei publishes blog calling for mandatory pre-release testing and government-backed independent auditing [22][23]
  • 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 [30]
  • 2026-06-14: David Sacks publicly accuses Anthropic of 'regulatory capture strategy based on fear-mongering'; Amodei responds defending his regulatory stance [38][24][40][41][42]
  • 2026-06-14: US government forces Anthropic to disable Fable-class AI models for all foreign nationals [6][43][7]
  • 2026-06-15: Florida expands its lawsuit against OpenAI to 83 pages naming CEO Sam Altman personally, separate from the ongoing criminal probe [31][67]
  • 2026-06-18: DOJ moves to protect xAI's Mississippi AI data center, treating high-speed AI compute as national security infrastructure [9]
  • 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 [26][21]
  • 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 Meta agreement is imminent [14][10]
  • 2026-06-27: GPT-5.6 Sol launches as a government-gated controlled preview; OpenAI publishes a system card under its Deployment Safety Hub [37][51][52][76][16][17]
  • 2026-06-27: Meta confirmed still holding out on government security reviews despite White House confidence that an agreement was imminent [12][13]
  • 2026-06-30: White House lifts export control on Anthropic, reversing the June 14 ban that blocked Fable-class models for foreign nationals; no public explanation given [8]

Perspectives

Trump administration (federal)

Signed a voluntary AI EO with NSA oversight disclaiming binding requirements; simultaneously imposed and then lifted export controls on Anthropic, 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 Sol.

Evolution: The June 30 reversal of Anthropic's export control — imposing and lifting the ban within two weeks without stated explanation — adds a new dimension: the administration can deploy and withdraw coercive controls outside the EO's stated terms, but whether this reflects enforcement capacity or improvised leverage is unclear.

[32][4][1][2][34][3][9][6][26][35][36][14][10][37][12][8]

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; the faction's influence is visible in the June 3 EO's narrow voluntary terms, though the administration's subsequent coercive actions sit in tension with the accelerationist position.

[4][5][38][24][25][39]

Anthropic / Dario Amodei

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

Evolution: Export controls requiring Fable-class models be disabled for foreign nationals were imposed June 14 and lifted June 30 without public explanation; whether Anthropic made concessions or the ban was temporary leverage is unknown from available reporting, complicating the characterization that Anthropic's principles make it a durable White House target.

[23][22][40][41][42][6][43][26][7][8]

OpenAI

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

Evolution: The Sol 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 post-deployment access management without statutory basis.

[44][45][46][47][48][49][50][30][14][37][51][52][16]

Meta

Refusing to join the voluntary pre-release review system despite administration pressure; has not complied as of early July.

Evolution: The White House expressed public confidence that an agreement was imminent, but Meta's continued holdout — confirmed by multiple reports through July 1 — shows that confidence was premature.

[35][53][54][36][55][56][57][58][10][59][12][13][11]

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.

[29][60][28][61][62][63][27]

State AGs (42-state coalition + Florida litigation)

Pursuing 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.

[64][65][66][30][31][67][68][69][70]

Eric Schmidt

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

Evolution: Consistent; supplies the most direct public justification 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 [22]; Sacks and Vance argue safety advocates manufacture fear for commercial reasons [24][25] — yet the administration simultaneously imposed and lifted export controls on Anthropic's Fable models within two weeks [6][8]. [22][24][6][25][8]
  • Federal preemption preference vs. state enforcement reality: Trump's preemption EO lacks constitutional authority absent congressional action [32][33]; binding Illinois law [27] and a 42-state coalition subpoena [30] advance without federal coordination. [32][33][27][30]
  • National security access control vs. risk of accelerating competitor adoption: the government gated GPT-5.6 Sol behind customer-by-customer approval citing cybersecurity [14][16], but critics argue this pushes developers toward Chinese and open-weight alternatives [19][20][71], working against the US compute-dominance rationale [21]. [14][16][19][20][71][21]
  • 'Voluntary' review framework vs. coercive application: the administration labeled the review system voluntary [1][3] but imposed and then reversed export controls on Anthropic [6][8], pressed Meta to join [10], and secured a government-gated Sol launch [37][51] — none of which is contemplated in the stated voluntary framework. [1][3][6][8][10][37][51]
  • OpenAI's governance advocacy vs. its PAC's political operations and legal exposure: OpenAI calls for democratic governance and shared evaluation standards [45] while Build American AI admitted running sockpuppet accounts [50] and the 42-state subpoena targets the consumer practices OpenAI presents as responsible [30]. [45][48][50][30]
  • Government access control vs. absence of statutory authority: the administration's role as access gatekeeper for GPT-5.6 Sol [37][16] and its use of export controls against Anthropic [6][8] extend executive oversight without clear statutory basis, beyond what the June 3 EO authorized [1][3]. [37][16][6][8][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] White House lifts export control on Anthropic that froze its most ... — reactive:us-ai-policy-regulation
  9. [9] Large AI campuses are indeed strategic infrastructure of a nation. — Rohan Paul Twitter (2026-06-18)
  10. [10] '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
  11. [11] The US is pressing Meta to let it review its AI, and Meta is the last holdout — reactive:us-ai-policy-regulation
  12. [12] Meta holds out on government security reviews of its AI models — reactive:us-ai-policy-regulation
  13. [13] US presses Meta on AI model review - LinkedIn — reactive:us-ai-policy-regulation
  14. [14] 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)
  15. [15] OpenAI Launches Limited Preview of GPT-5.6 at US Government's ... — reactive:gpt-56-launch-government-access
  16. [16] OpenAI Previews GPT-5.6 Sol With Restricted Access and Stronger ... — reactive:us-ai-policy-regulation
  17. [17] OpenAI Limits Access to New Models, Citing Government Security ... — reactive:us-ai-policy-regulation
  18. [18] When AI Safety Becomes AI Feudalism — reactive:us-ai-policy-regulation (2026-06-29)
  19. [19] If the U.S. turns frontier AI into a domestic-only or government-approved club, it risks doing to AI what it would never... — reactive:europe-ai-sovereignty-deficit (2026-06-27)
  20. [20] China Wins: Why Open-Source AI Becomes the Default — and Why $TAO Matters — reactive:us-ai-policy-regulation (2026-06-28)
  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] 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)
  23. [23] 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)
  24. [24] 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
  25. [25] 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)
  26. [26] Bloomberg, “Anthropic's AI Principles Make It a White House Target” — reactive:us-ai-policy-regulation
  27. [27] Illinois Signs Frontier AI Safety Bill into Law | Chris Lehane posted ... — reactive:us-ai-policy-regulation
  28. [28] 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
  29. [29] Bill Text: CA SB951 | 2025-2026 | Regular Session | Amended — reactive:us-ai-policy-regulation
  30. [30] 😺 42 states just subpoenaed OpenAI — The Neuron (2026-06-15)
  31. [31] Florida AG Sues OpenAI, Alleges ChatGPT Aided Mass Shooters — reactive:us-ai-policy-regulation
  32. [32] President Trump Signs Executive Order Preempting State AI Laws ... — reactive:us-ai-policy-regulation
  33. [33] President Trump Signs Executive Order Challenging State AI Laws | Paul Hastings LLP — reactive:us-ai-policy-regulation
  34. [34] Trump AI Executive Order Gives NSA Classified Role Over Frontier Models — reactive:us-ai-policy-regulation
  35. [35] NYT: The Trump administration is pressuring Meta to submit its AI models to government review before public release. — Rohan Paul Twitter (2026-06-24)
  36. [36] Meta Faces Growing Pressure From Trump Administration To Submit Advanced AI Models For Federal Testing: R - Benzinga — reactive:us-ai-policy-regulation
  37. [37] OpenAI Limits GPT-5.6 Rollout at US Government's Request — reactive:us-ai-policy-regulation
  38. [38] David Sacks is done being polite about Anthropic (Save this). — Milk Road AI Twitter (2026-06-14)
  39. [39] Sacks urges focus on Mythos-driven security needs, seeks to ‘correct’ the narrative around his Anthropic critique | InsideAIPolicy.com — reactive:us-ai-policy-regulation
  40. [40] Anthropic CEO disputes David Sacks' claims that company is 'woke' — reactive:us-ai-policy-regulation
  41. [41] Anthropic CEO defends support for AI regulations, alignment with ... — reactive:us-ai-policy-regulation
  42. [42] Anthropic CEO claps back after Trump officials accuse firm of AI fear-mongering   | TechCrunch — reactive:us-ai-policy-regulation
  43. [43] Legal Considerations Related to the Anthropic “Export Controls ... — reactive:us-ai-policy-regulation
  44. [44] OpenAI’s Frontier Governance Framework — OpenAI Blog (2026-05-28)
  45. [45] OpenAI public policy agenda — OpenAI Blog (2026-06-03)
  46. [46] A blueprint for democratic governance of frontier AI — OpenAI Blog (2026-06-03)
  47. [47] 🟡 Let’s get physical — Semafor Technology (2026-06-03)
  48. [48] AI #171: False Flag — Zvi's AI Roundups (2026-06-04)
  49. [49] The OpenAI-Andreessen-Palantir SuperPAC admits that it was “part ... — reactive:us-ai-policy-regulation
  50. [50] Build American AI admits hiring an outside vendor to run anonymous sockpuppet accounts targeting its opponents · Digg — reactive:us-ai-policy-regulation
  51. [51] OpenAI releases GPT-5.6 with government as gatekeeper - LinkedIn — reactive:us-ai-policy-regulation
  52. [52] GPT-5.6 Preview System Card - OpenAI Deployment Safety Hub — reactive:us-ai-policy-regulation
  53. [53] Meta Faces White House Push to Join AI Pre-Release ... — reactive:us-ai-policy-regulation
  54. [54] US Government Pressures Meta to Allow Pre-Release Reviews of Its ... — reactive:us-ai-policy-regulation
  55. [55] Trump administration pushes Meta for AI model security reviews- NYT — reactive:us-ai-policy-regulation
  56. [56] 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)
  57. [57] Meta Under Pressure to Join Federal AI Model Review Program - Techstrong.ai — reactive:us-ai-policy-regulation
  58. [58] Trump Administration Presses Meta To Submit AI Models For Federal Review — reactive:us-ai-policy-regulation
  59. [59] US Government Reportedly Urging Meta To Share Its AI Models — reactive:us-ai-policy-regulation
  60. [60] SB 951: Employment: technological displacement: notice. — reactive:us-ai-policy-regulation
  61. [61] California Set to Restrict AI Use in the Workplace With “No Robo ... — reactive:us-ai-policy-regulation
  62. [62] Trump loses more control over AI regulation as Illinois passes landmark law — Ars Technica AI (2026-05-28)
  63. [63] Illinois Legislature passes historic AI bill that would require third-party safety audits — reactive:us-ai-policy-regulation
  64. [64] ChatGPT creators knew product would cause harm, Florida argues in lawsuit • Florida Phoenix — reactive:us-ai-policy-regulation
  65. [65] Attorney General James Uthmeier Launches Criminal Investigation into OpenAI, ChatGPT | My Florida Legal — reactive:us-ai-policy-regulation
  66. [66] Florida AG brings 'first-in-the-nation state-led lawsuit' against OpenAI and its CEO — reactive:us-ai-policy-regulation
  67. [67] Florida Attorney General Sues OpenAI Over ChatGPT's Alleged ... — reactive:us-ai-policy-regulation
  68. [68] Florida attorney general sues OpenAI | Courthouse News Service — reactive:us-ai-policy-regulation
  69. [69] Florida sues OpenAI over ChatGPT risks Attorney General ... — reactive:us-ai-policy-regulation
  70. [70] AI and Liability — Simon Willison (2026-06-25)
  71. [71] 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)
  72. [72] President Trump Signs Executive Order to Block State AI Laws — reactive:us-ai-policy-regulation
  73. [73] WATCH: Trump explains why he postponed signing AI executive order — reactive:us-ai-policy-regulation
  74. [74] A Pro-AI Super PAC’s Secret Meme Sockpuppets — reactive:us-ai-policy-regulation
  75. [75] A Dark-Money Campaign Is Paying Influencers to Frame Chinese AI as a Threat | WIRED — reactive:us-ai-policy-regulation
  76. [76] [PDF] GPT-5.6 Preview System Card - Deployment Safety Hub — reactive:us-ai-policy-regulation