Hiring, screening, promotion, bias audits, notice, discrimination and recordkeeping.
AI regulation is now a jurisdiction problem.
A practical, source-backed map of U.S. federal AI policy, every state and D.C., significant local rules, current federal proposals, sector regulation and major international frameworks. Built for people who need to know what is binding, what is coming and where to verify it.
Complete jurisdiction coverage without pretending a static page is a legislature.
The state section covers all 50 states plus D.C. directly. For bill-level completeness, each card links to a live state inventory and to NCSL’s Artificial Intelligence Legislation Database, which tracks introduced state AI legislation from 2025 forward. Key laws in force are cross-checked against primary sources and a monthly defined-law tracker.
Chatbots, companions, disclosures, deceptive practices, privacy, automated decisions and redress.
Clinical software, insurer utilization management, mental-health bots, patient communications and human oversight.
Deepfakes, intimate imagery, election content, provenance, voice/likeness and labeling.
Consequential decisions, impact assessments, risk management, appeals and human review.
Age assurance, companion bots, manipulative design, safety protocols and sexual content restrictions.
Safety frameworks, incident reporting, systemic-risk controls, model transparency and whistleblower protections.
Agency inventories, federal acquisition rules, vendor controls, public-sector transparency and prohibited uses.
Federal AI policy is a stack, not one AI Act.
There is no single comprehensive U.S. AI statute. The current national compliance picture is distributed across federal law, executive policy, agency use and procurement rules, existing regulator authority, technical standards and sector-specific requirements.
All 25 bills in the current federal AI monitor.
These are the AI-related measures currently shown in a live Congress.gov-derived tracker as of September 16, 2026, ordered by latest legislative activity. They are proposals unless and until enacted.
| Bill | Subject / title | Latest status | Latest action | Official |
|---|
Every state + D.C.
Search the current landscape by state, region or issue. Each card summarizes the practical themes, shows the September 16 published-item snapshot and links to the live bill record, NCSL and the defined-law tracker.
Rules a business is more likely to feel.
This is not every enacted AI statute. It highlights high-impact requirements and dates that help translate the broader state inventory into operational planning.
| Jurisdiction | Law / rule | Status | Timing | Scope | Operational takeaway |
|---|
City rules can create a second compliance layer.
Local rules are not uniform. These examples show why state-only research can miss hiring, facial-recognition, procurement and government-use requirements.
Major global approaches are diverging.
For companies with users, workers, vendors or products outside the U.S., the important question is not whether another country has an “AI Act.” The important question is which binding laws, sector rules, guidance and extraterritorial duties can reach your use case.
Turn the regulatory map into an AI governance process.
The fastest way to get useful compliance visibility is to map AI systems to people, decisions and jurisdictions, then attach the right controls to the highest-risk uses.
Include purchased SaaS, embedded vendor features, internally built models, chatbots, scoring tools, content generators and AI agents.
Record where employees, applicants, customers, patients, students and affected consumers are located, not just where your company is incorporated.
Flag employment, credit, housing, insurance, healthcare, education, government benefits, biometrics, children, companions, synthetic media and frontier-model activity.
Developer, provider, deployer, employer, platform, data broker, government contractor and downstream user can carry very different duties.
Where required or prudent, document disclosure language, escalation, adverse-decision explanations, appeals and meaningful human review.
Keep model/vendor documentation, data lineage, testing, impact assessments, audits, incident records, approvals and policy exceptions.
Require cooperation on regulatory disclosures, incident reporting, model changes, data rights, security, audit evidence and exit/portability.
Separate “proposed,” “enacted but upcoming,” and “in force.” Recheck high-activity jurisdictions monthly or before a material AI launch.
Use primary sources first.
This atlas is a navigation and research tool, not legal advice. Source priority is official law or regulator, then NCSL or an equivalent legislative source, then secondary trackers for discovery and cross-jurisdiction comparison.