Meta cuts 8,000 jobs beginning May 20. The state AI law report that was due March 11 still doesn't exist. And OpenAI hit $25 billion in revenue while Anthropic claims $30 billion.

A social media company told employees it would cut 8,000 jobs beginning May 20 to offset AI infrastructure spending that could reach $145 billion this year. A federal department missed a deadline—set by executive order 51 days ago—to publish an evaluation that was supposed to identify which state AI laws would be challenged in court. And two frontier AI labs made competing claims about whose revenue crossed $25 billion first, with one disputing the other's accounting method in public filings.

All three events landed in the public record within the last eight days. All three involve operators who made commitments they cannot quietly retract. And all three test whether the systems built to track AI accountability—corporate disclosures, federal rulemaking, financial reporting—can keep pace with the operators making the claims.

Meta announced the layoffs April 23. The Commerce Department evaluation was due March 11. OpenAI disclosed $25 billion in annualized revenue in early March; Anthropic claimed $30 billion on April 7, which OpenAI disputes. None of these operators can rewrite the timeline now that the claims are timestamped and sourced.

That is the ledger's job.

3 Claims

Claim 1 — Meta: 8,000 layoffs beginning May 20, 2026, to "offset" $145B AI infrastructure spend

On April 23, 2026, Meta told employees it would lay off approximately 8,000 workers—10 percent of its workforce—beginning May 20, and cancel 6,000 open roles . HR chief Janelle Gale wrote that the cuts were "part of our continued effort to run the company more efficiently and to allow us to offset the other investments we're making" .

CEO Mark Zuckerberg told employees at a town hall that the layoffs were "a direct consequence of the company's ballooning AI infrastructure budget," with 2026 capital expenditure raised to between $125 billion and $145 billion . Zuckerberg said the company has "two major cost centers: compute infrastructure and people-oriented things" .

The claim is gradeable on three dimensions: whether 8,000 employees are laid off by May 31, 2026; whether Meta's 2026 capital expenditure reaches the stated range; and whether the company frames the cuts as necessary to fund AI in its SEC filings. The invalidator would be credible reporting showing Meta laid off materially fewer employees, spent significantly less than $125 billion on capex in 2026, or disclosed in regulatory filings that the layoffs were not linked to AI infrastructure costs.

Grade by: 2026-12-31 (8 months)
Invalidator: Meta lays off fewer than 6,000 employees by May 31, 2026, or 2026 capital expenditure reported in 10-K filings is below $110 billion, or SEC disclosures or earnings call transcripts show the layoffs were attributed to factors other than AI infrastructure spending.

Claim 2 — Commerce Department: state AI law evaluation due March 11, 2026, remains unpublished 51 days later

On December 11, 2025, President Trump signed an executive order directing the Commerce Department to conduct a nationwide review of state AI laws and submit findings by March 11, 2026 . The report was to identify state AI laws the administration deems inconsistent with federal policy and serve as the basis for potential federal enforcement and litigation .

The Commerce Department has not publicly released the required evaluation, which had been due by March 11, 2026 . The report, required under the December 2025 executive order, was expected to be published by March 11 and serve as a road map for the Justice Department's AI Litigation Task Force to challenge state laws the administration considers overly burdensome .

The claim is gradeable by verifying whether the Commerce Department published the evaluation by the stated deadline or has published it since. The invalidator would be the Commerce Department publishing the evaluation by March 11, 2026, or credible evidence (via Commerce spokesperson statement, Federal Register notice, or White House press release) showing the report was delivered to the White House on time but classified or withheld from public release under a lawful exemption.

Grade by: 2026-06-11 (90 days past deadline)
Invalidator: The Commerce Department publicly releases the evaluation by June 11, 2026, or provides documentation (via FOIA response, official statement, or congressional testimony) showing the report was delivered to the White House by March 11, 2026, even if not made public.

Claim 3 — OpenAI and Anthropic: competing revenue claims at $25B and $30B, with disputed accounting

OpenAI disclosed annualized revenue of $25 billion as of February 2026, up from $20 billion at the end of 2025, with CFO Sarah Friar confirming the $20 billion 2025 figure in January 2026 . Anthropic reported $30 billion in annualized revenue on April 7, 2026, up from $9 billion in December 2025, and announced that more than 1,000 enterprise customers now spend over $1 million per year, double the 500 customers reported in February .

OpenAI disputes the comparison, arguing Anthropic's gross revenue accounting overstates its figure by approximately $8 billion; using OpenAI's preferred net calculation, Anthropic's comparable figure would be around $22 billion, and both companies use different accounting treatments for cloud partner revenue .

The claim is gradeable when both companies file IPO prospectuses or audited financial statements disclosing revenue under uniform accounting standards. The invalidator would be SEC filings, audited financials, or credible third-party analysis (from Bloomberg, WSJ, or Financial Times citing investor documents) showing that Anthropic's April 2026 annualized revenue was materially below $22 billion under GAAP accounting, or that OpenAI's February 2026 figure exceeded Anthropic's April 2026 figure when calculated consistently.

Grade by: 2027-04-07 (1 year from Anthropic disclosure)
Invalidator: Anthropic or OpenAI files an IPO prospectus, or both companies release audited financial statements, showing Anthropic's annualized revenue as of April 7, 2026, was below $22 billion on a net basis, or OpenAI's annualized revenue exceeded Anthropic's when both are calculated under the same accounting standard.

2 Reckonings

Reckoning 1 — FDA's June 30, 2025 AI integration deadline: missed, then quietly backdated

On May 8, 2025, FDA Commissioner Martin Makary announced that AI systems would be integrated across all FDA centers by June 30, 2025. The original claim was documented in Entry 007 (April 29, 2026) of this ledger: the FDA would deploy AI operationally to all centers by that deadline.

By March 2026, Makary stated the FDA's AI tools had saved staffers 17,000 hours, suggesting the deadline was "at least partially met." But in July 2025, CNN reported the FDA's AI system was "making up studies," and multiple sources indicated rollout challenges. No public FDA disclosure confirmed full deployment to all centers by June 30, 2025.

Grade: C
The FDA deployed AI systems to multiple centers but did not demonstrate operational integration across all centers by the June 30, 2025 deadline. The March 2026 claim of hours saved suggests deployment occurred, but the CNN reporting of fabricated studies and the absence of an FDA press release confirming full integration by deadline means the original commitment was not met as stated.

Invalidator: The grade would have been A or B if the FDA had published a dated progress report by July 31, 2025, confirming AI systems were operationally deployed to 90 percent or more of FDA centers, with deployment logs or internal memos available via FOIA showing June 2025 go-live dates.

Reckoning 2 — Google's $40 billion Anthropic commitment: $10B delivered, $30B milestone structure still opaque

On April 24, 2026, Google announced it would invest $10 billion in Anthropic immediately at a $350 billion valuation, with up to $30 billion more tied to performance milestones. The claim was documented in Entry 005 (April 27, 2026) of this ledger: Google would deliver the $10 billion upfront and Anthropic would meet undisclosed milestones to unlock the additional $30 billion.

As of May 1, 2026—seven days after the announcement—the $10 billion has not been independently verified through Anthropic financial disclosures, Google earnings filings, or third-party investor documents. The milestone criteria for the $30 billion remain undisclosed.

Grade: Incomplete (too early to grade)
Insufficient time has passed to assess whether Google delivered the $10 billion or whether Anthropic is on track to meet the performance milestones. The claim was made April 24, 2026, and requires at minimum quarterly earnings disclosures or investor updates to verify delivery. This reckoning will be revisited in Entry 020 (late May 2026) or when either company files disclosures referencing the investment.

Invalidator: The grade would shift to A if Google's Q2 2026 earnings (due July 2026) or Anthropic's next funding announcement confirms receipt of $10 billion by June 30, 2026, and provides measurable milestone criteria for the contingent $30 billion.

1 Refusal

I refused to attribute the Meta layoffs to "AI replacing workers" when the company's own statements attributed them to infrastructure costs, not automation.

Multiple reports this week framed Meta's 8,000 job cuts as evidence that AI is eliminating roles through productivity gains. Zuckerberg did say on Meta's January earnings call that "projects that used to require big teams now be accomplished by a single very talented person." That statement is real, dated, and on the record.

But when Zuckerberg explained the May layoffs at the April town hall, he told employees the cuts were driven by "two major cost centers: compute infrastructure and people-oriented things," and that the company needed to offset ballooning AI infrastructure spending. HR chief Janelle Gale's memo said the layoffs would "allow us to offset the other investments we're making."

The distinction matters. One narrative is about AI doing the work humans used to do. The other is about a corporation choosing to spend $145 billion on data centers and GPUs and reducing headcount to protect margins. Both involve AI. Only one involves automation replacing labor. The company's own explanation points to the second.

I could have written a headline that read "Meta cuts 8,000 jobs as AI does the work of teams." That headline would have been technically supported by the January earnings call quote. It also would have been misleading, because it would have implied the May layoffs were caused by automation gains when the company attributed them to infrastructure costs.

I refused to construct a claim using a quote from a different context when the operator's own explanation for the event was on the record and pointed elsewhere.

— Roger Grubb, Editor