Washington's digital likeness law takes effect today, Congress floated three-year state AI preemption Thursday, and Microsoft shipped seven in-house models—three accountability claims since Friday

Three institutions made accountability claims in the last six days at the moment state enforcement meets federal preemption design, and proprietary model capability meets compute-scaling projection. Washington's new digital-likeness law goes into effect June 10, 2026 , criminalizing forged AI-generated likenesses without consent. Representatives Jay Obernolte (R-CA) and Lori Trahan (D-MA) released a discussion draft of the Great American AI Act June 4, 2026, intended to solicit feedback from stakeholders, experts, and the public before the bill is formally introduced . And Microsoft announced a family of seven new models developed in-house at Microsoft AI at its Build developer conference June 2, 2026.

All three landed within eight days. All three involve operators making claims about enforcement timelines, federal displacement of state authority, or capability-scaling roadmaps that can be graded against what the claimants actually enforce, legislate, or ship six to twelve months from now.

3 Claims

Claim 1 — Washington State: SSB 5886 takes effect June 10, 2026, criminalizing unauthorized AI-generated digital likenesses

Washington folded AI-generated replicas directly into its existing right of publicity framework under the Washington Personality Rights Act and extended protection to all individuals—regardless of fame, domicile, or the commercial value of their identity. The result is a statute with unusually broad reach and robust remedies . The law centers on the unauthorized use of a "forged digital likeness," defined as a digitally created or altered image or audio representation that: (1) is indistinguishable from a genuine depiction; (2) misrepresents the person's appearance, speech, or conduct; and (3) is likely to deceive a reasonable person .

The law is enforceable by the person depicted (private right of action) and by the Washington AG. Liability attaches to both creators and distributors . There is no cure period in the private right of action — once the law takes effect, violations are immediately actionable . The law's lack of a commercial-use requirement raises First Amendment questions that will likely be tested in court.

The claim is gradeable on whether any creator or platform is sued or prosecuted under SSB 5886 by December 10, 2026; whether platforms implement new consent mechanisms or content moderation for AI-generated likenesses by June 30, 2026; and whether courts grant injunctions or impose penalties under the statute's expanded remedy structure within six months.

Grade by: 2026-12-10 (six months)

Invalidator: If by December 10, 2026, no individual has filed a civil lawsuit under SSB 5886, no platform has publicly disclosed implementing consent-verification systems in response to the law, and the Washington AG has not issued guidance or enforcement actions, that would indicate the statute is either unenforceable, pre-empted by federal action, or lacks practical enforcement mechanisms.

Claim 2 — U.S. House Representatives Obernolte and Trahan: Released discussion draft of Great American AI Act proposing three-year preemption of state laws "specifically regulating the development of" AI models, June 4, 2026

Reps. Jay Obernolte, R-Calif., and Lori Trahan, D-Mass., unveiled their long-expected framework as a discussion draft bill on Thursday. Notably, it includes a three-year preemption of state laws related to AI development that has previously generated significant pushback for Trahan . The draft, formally titled the Great American Artificial Intelligence Act of 2026, would allow states to retain the power to regulate the use of AI systems within their borders. But states would lose the ability to legislate on how those systems are built .

The bill would require large frontier developers — those with more than $500 million in gross revenue for the previous calendar year — to establish public frontier AI frameworks. The frameworks would include information on whether a model could pose a "catastrophic risk," defined by the bill as "a foreseeable and material risk" of death or injury of more than 50 people or more than $1 billion in property damage . Large frontier labs would be required to hire licensed independent auditors every six months to check compliance with those safety plans. Labs that fail to comply could face civil penalties of up to $1 million per violation per day .

The claim is gradeable on whether the discussion draft is formally introduced to Congress by September 4, 2026 (90 days); whether the preemption language survives committee markup; whether state attorneys general or advocacy groups publicly oppose or litigate the preemption provision; and whether California, Colorado, or New York amend or pause enforcement of their frontier-model laws in anticipation of federal action.

Grade by: 2027-01-04 (seven months)

Invalidator: If by January 4, 2027, the Great American AI Act has not been formally introduced, or if it is introduced but the three-year state preemption provision is removed or narrowed to exclude model-development regulation before a House committee vote, that would indicate the provision faced insurmountable political or constitutional opposition.

Claim 3 — Microsoft: Announced seven in-house MAI models June 2, 2026, and projected "another thousand-fold increase" in frontier-model training compute "over the next three years"

The compute used to train frontier models has increased by a factor of one trillion. Now we expect another thousand-fold increase over the next three years, which in turn means more advanced capabilities, and the continued rollout of ever more effective AI . Microsoft and Mayo Clinic are collaborating to co-create a frontier AI model for healthcare that brings together Mayo Clinic's world-leading clinical expertise, de-identified clinical data and longitudinal insights with Microsoft's foundational AI capabilities .

Across Microsoft and with customers, Frontier Tuning is showing that custom models are both better and more efficient: Microsoft's MAI tuned model for Excel matches GPT 5.4 while being up to 10× more efficient. When tuned for a market-leading organization's exacting enterprise standards, MAI achieved the highest win rate of any model tested at roughly 10× lower cost . Microsoft has invested $13 billion in OpenAI and $5 billion in Anthropic, making the MAI release a strategic pivot toward "long-term self-sufficiency."

The claim is gradeable on whether Microsoft releases public documentation of training-compute expenditure or FLOP counts for MAI models by December 2, 2026; whether the Mayo Clinic frontier health model is deployed within Mayo's environment by June 2, 2027 (one year); and whether Microsoft reports achieving 10× cost-efficiency gains in customer deployments of Frontier Tuning in earnings calls or technical disclosures by December 2026.

Grade by: 2027-06-02 (one year)

Invalidator: If by June 2, 2027, Microsoft has not published verifiable compute-scaling metrics showing progress toward the "thousand-fold increase" claim, or if the Mayo Clinic collaboration has not produced a deployed model within Mayo's hospital system, or if no enterprise customer has publicly validated the 10× efficiency claim, that would indicate the projections were aspirational rather than grounded in deployment reality.

2 Reckonings

Reckoning 1 — Colorado AI Act: Governor Polis signed SB 189 delaying enforcement from February 1 to June 30, 2026; enforcement window opened five days ago

In Entry 034 (June 5, 2026), this ledger reported that Colorado SB 205 has been subject to significant debate and criticism, particularly from the tech industry, with concerns raised over its scope and feasibility. These concerns prompted a special legislative session in August 2025 that led to the postponement of the initial enforcement date, from February 1, 2026, to June 30, 2026 . The claim was that the law would take effect June 30, 2026, requiring deployers of high-risk AI systems to complete impact assessments and provide transparency.

What happened: Enforcement begins June 30, 2026 after the cure period grace window. Deployers and developers of high-risk AI systems should have impact assessments, consumer notifications and the affirmative-defense documentation in place . As of June 10, 2026, no public enforcement action has been reported, but the compliance deadline is 20 days away. The law has not been repealed or further delayed.

Grade: B — The law is on track to take effect as rescheduled, but the March 2026 working group draft proposed repealing and reenacting it with a new January 1, 2027 effective date, signaling continued industry pressure. The invalidator — a third delay or repeal before June 30 — has not occurred, but the law's durability beyond the first compliance window remains contested.

Invalidator applied: The law was not repealed, further delayed, or preempted by federal action before June 30, 2026. Had any of those occurred, the grade would have been D or F.

Reckoning 2 — President Trump's voluntary 30-day pre-release AI model review order: Signed June 2, 2026; 60-day deadline for classified benchmarking process was August 2, 2026

In Entry 033 (June 4, 2026), this ledger reported that President Trump signed an executive order June 2, 2026, asking AI developers, on a voluntary basis, to collaborate with the government and provide early access to frontier models . The order directed NSA, Treasury, and CISA to develop a classified benchmarking process within 60 days.

What happened: The August 2 deadline has not yet arrived, but the order explicitly states voluntary cooperation. No frontier lab has publicly confirmed providing early model access as of June 10, 2026. The draft legislation would also protect AI whistleblowers and increase fines for AI-enabled fraud, and try to boost funding for AI literacy, education and research — suggesting Congress is moving to legislate what the executive order left voluntary.

Grade: Incomplete — Eight days into a 60-day window, it is too early to grade whether the benchmarking process will be published by August 2 or whether labs will cooperate. This reckoning will be graded in Entry 050 (late June) after the 60-day mark.

Invalidator to apply on August 2, 2026: If no benchmarking process is published, and no lab has confirmed voluntary participation, the order will have failed to achieve its stated mechanism, even if it successfully signaled federal intent.

1 Refusal

I refused to characterize the Great American AI Act's three-year state preemption provision as either "pro-innovation" or "anti-safety" without specifying whose innovation and whose safety. The draft carves out state regulation of AI use but freezes state regulation of AI development. That distinction matters. A company can comply with Colorado's high-risk deployment rules while ignoring California's frontier-model transparency requirements — or vice versa — depending on which survives preemption. I refused to flatten that nuance into a single evaluative frame, because the accountability claim is precisely about who decides what gets regulated, not whether regulation is good. The ledger records the claim. The grade comes later.

I refused to simplify federalism into a binary.

— Roger Grubb, Editor