Big Tech Spent $150 Million to Write America's AI Rules
Lobbying disclosures, White House policy documents, and campaign finance records trace a direct line from corporate treasuries to federal AI regulations that systematically neutralize state oversight.
The single most documented fact in the AI regulatory debate is this: a $150 million lobbying campaign — catalogued by Forbes on November 28, 2025 — preceded and directly corresponded with the White House's December 11, 2025 National AI Policy Framework, a document that moved to establish federal supremacy over state AI regulation at the precise moment state legislatures were advancing their own accountability measures. That is not coincidence. That is a money trail with a paper destination.
The financial architecture behind that outcome is now visible in mandatory federal filings. Under the Lobbying Disclosure Act, 2 U.S.C. § 1604, companies must report quarterly lobbying expenditures to the Secretary of the Senate and Clerk of the House. Those filings, analyzed by the nonpartisan watchdog Issue One, show that six companies — Alphabet, Anthropic, Meta, Microsoft, Nvidia, and OpenAI — collectively maintained 324 registered lobbyists in Q2 2026 alone, spending a combined rate exceeding $226,000 per day on federal influence operations. By Issue One's most recent calculation, that figure had risen to $230,000 per day. For context, a Q3 2025 Issue One analysis of seven major tech companies found combined federal lobbying spend of $50 million in that period alone.
The company-level LDA filings reveal a telling acceleration pattern. Anthropic's quarterly lobbying expenditure rose from $360,000 in Q1 2025 to $1,560,000 in Q1 2026 — a 333 percent increase in twelve months, reported by Fortune on April 23, 2026. Fortune specifically noted this surge occurred against the backdrop of a 'rocky relationship with the White House,' meaning the company quadrupled its lobbying investment precisely when organic access had deteriorated. OpenAI's quarterly spend rose from $560,000 to $1,020,000 over the same period — an 82 percent increase — as it simultaneously transitioned to a for-profit structure and pursued federal AI procurement contracts. The money moved in the same direction as the policy need.
The policy outputs that followed this investment are documented in public records. The Trump administration's AI Action Plan, released July 2025, contained four provisions with direct alignment to the lobbying priorities these companies had funded. The plan fast-tracked AI data center permitting, encouraged replacement of federal government functions with AI products, weakened oversight requirements for AI procurement, and — most consequentially — established a federal funding conditionality mechanism requiring agencies to 'consider a state's AI regulatory climate when making funding decisions and limit funding if the state's AI regulatory regimes may hinder the effectiveness of that funding or award.' A Congress.gov document from the House Judiciary Committee [HHRG-119-JU05-20251216-SD014-U14.pdf] explicitly characterized this provision as a 'recapitulation of the AI regulation moratorium,' connecting it to prior legislative efforts to preempt state-level AI accountability laws.
The strategic logic of preemption is documented by the Dutch research organization SOMO, which analyzed the industry's regulatory influence campaign. SOMO records that the industry deployed a specific narrative frame — 'Keep California Leading the AI Race' — to recast regulatory opposition as economic patriotism, tying corporate interests to regional economic identity. The goal, documented across the Forbes lobbying analysis, the SOMO report, and the Balsillie School of International Affairs research, was a federal framework that would neutralize California, Colorado, Illinois, and other states then advancing stricter AI consumer protection laws. According to SOMO's findings, at the federal level, Big Tech's anti-regulation lobbying 'has worked,' with the AI Action Plan 'openly prioritising stripping away what it calls red tape or onerous regulation.'
The public-private divergence at the center of this influence operation is itself documented. TIME magazine, citing congressional staffers with direct knowledge and nonprofit advocates who participated in legislative proceedings, reported that while Big Tech companies publicly expressed support for AI regulation in official testimony, in closed-door conversations with officials they 'consistently pushed for light-touch and voluntary rules.' That gap between public position and private lobbying is not an allegation — it is the documented account of people present in both rooms. The staffers who confirmed it are unnamed under standard source protection practices, but their accounts are corroborated by the policy outcomes the filings describe.
The revolving door mechanism that delivered access for these lobbying dollars is documented in principle if not yet fully mapped in individual names. Public Citizen has documented that Big Tech lobbying teams systematically recruit former congressional staffers, former FTC officials, and former personnel from NTIA, Commerce, and the DOJ Antitrust Division — individuals who bring not just expertise but personal relationships with sitting officials. This practice is legal under 18 U.S.C. § 207's cooling-off provisions, but it represents a structural conversion of public-sector tenure into private-sector access. The 324 registered lobbyists in Q2 2026 LDA filings represent only those meeting the statutory 20 percent time threshold; strategic advisors, trade association representatives, and law firm partners working below registration thresholds are not counted.
What remains hidden is substantial. The specific White House officials who drafted the July 2025 AI Action Plan and the December 2025 National AI Policy Framework, and their prior professional relationships with the companies whose interests those documents advance, are not yet in the public record. The trade association layer — including the Computer and Communications Industry Association, TechNet, the Information Technology Industry Council, and Chamber of Commerce technology divisions — allows member companies to fund lobbying that does not appear in individual LDA disclosures, representing a documented gap in the money trail. The full breakdown of which companies contributed to the $150 million preemption lobbying figure requires direct review of the Forbes November 28, 2025 reporting and corresponding LDA database records. The instruments that would close these gaps are available: a Government Accountability Office audit of revolving door compliance for AI policy personnel under 18 U.S.C. § 207, a cross-referenced analysis of LDA filings against White House visitor logs under Freedom of Information Act requests, and systematic review of trade association lobbying disclosures filed separately from member company reports. The financial architecture is documented. The policy outcomes are documented. The mechanism connecting them is partially visible. What remains in the dark is who, specifically, sat across the table when the rules were written — and what they were paid before they took that seat.