AI
The AI Lobby

Methodology

State Regulation Scores (1–10)

Each state receives a regulation score from 1 to 10 based on four weighted factors:

  • 20%Number of AI bills introduced
  • 30%Bills enacted into law
  • 20%Enforcement actions taken
  • 30%Regulatory approach — proactive vs. reactive

A score of 10 means comprehensive AI legislation with active enforcement. A score of 1 means no AI-specific legislation.

What Counts as an “AI Bill”

We search legislative databases for the following keywords: “artificial intelligence,” “machine learning,” “algorithmic,” “automated decision,” “deepfake,” “autonomous systems,” “generative AI,” and “large language model.”

Each result is manually reviewed to confirm AI relevance and categorized by topic — including privacy, employment, healthcare, elections, and more.

Lobbying Spending Attribution

Federal lobbying figures come from Senate LDA quarterly filings. We identify AI-related lobbying through a three-step process:

  1. Matching issue codes (SCI, CPT, TEC)
  2. Scanning the specific lobbying issues text field for AI-related keywords
  3. Attributing the full filing amount when AI is a listed issue (standard methodology, same as OpenSecrets)

Note: This may overcount when AI is one of many issues in a filing.

Influence Score (0–100)

A composite score computed from six weighted factors:

  • 30%Federal lobbying spending
  • 20%PAC contributions to relevant committee members
  • 15%Enforcement actions and fines
  • 15%Revolving door hires — former government officials
  • 10%Public comment submissions
  • 10%State-level lobbying where available

Data Sources

  • •Senate LDA API (lda.senate.gov)
  • •State legislature trackers (NCSL, MultiState, LegiScan)
  • •FEC filings (fec.gov)
  • •FTC enforcement records
  • •Brookings Institution
  • •National Conference of State Legislatures

Update Frequency

State billsWeekly
Federal lobbyingQuarterly (within 2 weeks of filing deadline)
Campaign contributionsMonthly
Company profilesAs needed

Limitations

  • •State lobbying data is incomplete — only ~30 states have machine-readable lobbying databases.
  • •We cannot attribute spending to specific AI issues within multi-issue lobbying filings.
  • •PAC data has a 30–60 day lag from FEC.
  • •Not all AI-related government activity involves the keywords we track.
About UsCorrections