For News, Media & Content Creators
Free access to our full database of anti-science legislation, sourced from LegiScan, GovInfo, and Open States. Browse, filter, bookmark, and share.
Quick Access
FILTERS SAVED IN THE URLAll Anti-Science Bills
Every bill we're tracking that threatens public health, science education, or evidence-based policy.
Browse all anti-science bills →High Priority / Active Threats
Bills actively moving through legislatures that pose the greatest risk right now.
View high-priority active bills →Pro-Science Legislation
Bills that strengthen public health protections, fund research, or support evidence-based policy.
Browse pro-science bills →Bills Signed into Law
Anti-science legislation that has already been signed. Track what passed and where.
View signed bills →Defeated / Dead Bills
Harmful bills that were stopped. Evidence that advocacy works.
View defeated bills →All Legislation
The full, unfiltered database. Apply your own filters and bookmark the result.
Browse all bills →Browse by Category
Covering a specific topic? Jump straight to the bills that matter for your story.
Browse by State
Covering your state legislature? See every anti-science bill in your state.
Bookmark & Share
How it works: When you apply filters on the Browse Bills page, the URL updates automatically. Copy the URL from your browser bar to save your exact filter setup. Share it with colleagues, embed it in articles, or bookmark it for future reference.
Example URLs:
tracker.html?state=TX&type=anti&status=active#browse-bills ← Active anti-science bills in Texas
tracker.html?category=vaccine-exemption&priority=High#browse-bills ← High-priority vaccine exemption bills
tracker.html?q=fluoride#browse-bills ← All bills mentioning fluoride
Our Methodology
KEYWORDS + AI, VERIFIED DAILYTransparency matters. Here's exactly how we find, classify, and verify every bill in our database.
Finding Bills
Every day, our automated system pulls newly introduced and updated legislation from three official sources:
- LegiScan - Covers all 50 state legislatures with real-time bill tracking
- GovInfo - The U.S. Government Publishing Office's official database for federal bills
- OpenStates - An open-source backup that ensures we don't miss anything
Our sources reach all 50 state legislatures and Congress. The live counts of bills and states currently in the database appear at the top of the homepage tracker. The system checks for new bills and status changes automatically every day.
First Pass: Keyword Analysis
Each bill's title, summary, and subject tags are scanned against a carefully curated set of topic keywords. A bill must first match at least one core topic we track:
Bills that don't mention any of these topics are classified as "Monitor" and kept in the database but not flagged.
For relevant bills, the system counts how many anti-science indicators vs. pro-science indicators appear in the text. For example:
Anti-Science Indicators
"religious exemption," "philosophical exemption," "parental rights" (in vaccine context), "medical freedom," "informed consent" (in vaccine context), "ban mandates," "gene therapy" (referring to mRNA), "raw milk sales," "remove fluoride"
Pro-Science Indicators
"strengthen immunization," "limit exemptions," "expand vaccine access," "fund public health," "evidence-based," "disease surveillance," "immunization registry," "pandemic preparedness"
If pro-science indicators outweigh anti-science ones, the bill is initially classified as "Pro-Science." If anti-science indicators are present, it's classified as "Anti-Science." If neither side is strong enough, it stays as "Monitor."
Second Pass: AI Verification
Keywords alone aren't enough. A bill titled "Eliminating Personal Conviction Exemption from Immunizations" contains the word "exemption" - but it's actually removing exemptions, which is pro-science. Simple keyword matching would get this wrong.
That's why every bill classified as anti-science or pro-science goes through a second, independent review using a large language model (AI). This AI reviewer:
- Reads the full context of each bill's title, summary, and category
- Makes its own independent classification without knowing what the keyword system decided
- Assigns a confidence score from 0% to 100% indicating how certain it is
- Cites specific evidence - the exact phrases that drove its decision
Resolving Disagreements
When the keyword system and the AI reviewer disagree, or when confidence is low, we apply strict rules:
Every disagreement is logged with the evidence from both systems. We regularly review these logs to improve both the keyword lists and the AI prompts.
Category Assignment
Beyond anti/pro classification, each bill is placed into a specific topic category based on its content:
This lets journalists and researchers quickly find bills relevant to their specific story or area of coverage.
Ongoing Updates
Our system runs automatically every day. When a bill's status changes - moves to a new committee, gets amended, passes a chamber, or is signed into law - the tracker updates within 24 hours. New bills are picked up as soon as they're filed.
All of our code is open source on GitHub under the GPLv3 license. Anyone can inspect, audit, or improve our classification algorithms.
Data Sources & Materials
Every number on this site comes from an official, verifiable source. Here's where it all originates.
Legislation Data
LegiScan API - Primary source for all 50 state legislatures. Provides bill text, status updates, sponsors, committee assignments, roll call votes, and subject tags. Updated daily. 30,000 requests/month.
GovInfo (GPO) - Official U.S. Government Publishing Office. Source for all federal bills (House & Senate). Free, no rate limits.
OpenStates API - Open-source backup for state legislation. Ensures coverage if LegiScan misses a bill or has delayed updates.
Legislator & Contact Data
Google Civic Information API - When you enter your address, this official Google service identifies your exact federal, state, and local representatives based on your voting districts.
OpenStates People API - Provides state legislator contact details including office emails, phone numbers, and committee memberships.
Email addresses and phone numbers come directly from official government websites and legislative databases.
AI Classification
Claude Sonnet, run from Claude Code on our own machine - The AI model used for independent bill classification review. Bills in science-relevant categories that the keyword pass could not call are reviewed in batches; every proposed pro- or anti-science label is re-checked by a second pass instructed to refute it, and only labels that survive at 0.70 confidence or higher are applied. Each reviewed bill's verdict, confidence, and verbatim evidence are published in the bill data.
The AI never invents information - it only evaluates the bill data already collected from official sources above.
Action Counts
Emails & calls tracked: When you use our "Email Your Rep" or "Call" tools, the action is counted via our Firebase backend. We count the action type (email or call) and nothing else - no personal data, no message content, no IP addresses.
Pledge data: Public officials who take the SAFE Action pledge are verified through our candidate quiz submission process and reviewed before publication.
Hosting & Infrastructure
Firebase Hosting - The website is a static site served by Firebase Hosting, with a small set of Cloud Functions for the mailing list and volunteer forms. No advertising trackers.
Firebase (Google Cloud) - Cloud functions handle action counting, volunteer applications, and donation processing. Firestore database stores bill data and aggregate counters.
GitHub Actions - Automated daily crawler runs at 6 AM UTC, pulling fresh data from all legislation APIs.
Open Source
Our entire codebase - including every keyword list, every AI prompt, every classification algorithm, and this website - is publicly available under the GPLv3 license.
Anyone can audit our methodology, file issues, or submit improvements.
A note on accuracy: No automated system is perfect. Our dual-layer approach (keywords + AI verification) with a conservative confidence threshold means we prioritize precision over recall - we'd rather leave a borderline bill as "Monitor" than risk mislabeling it. If you believe a bill has been misclassified, please open an issue on GitHub and we'll review it.