Back to all issues
Government

AI Governance & Regulation

Artificial intelligence is reshaping the economy, national security, and daily life faster than government can write rules—forcing hard choices about how, and how much, to regulate

The challenge
What's the Challenge?

Artificial intelligence is advancing faster than U.S. law and regulation. The technology touches almost every part of life—healthcare diagnoses, hiring decisions, lending, education, child safety, copyright, elections, and national security—and policy is being made in pieces: federal executive orders, state laws (notably in California, Colorado, and New York), agency guidance, and industry self-regulation. The European Union's AI Act takes a comprehensive risk-based approach; China is pursuing both heavy regulation and aggressive state investment. The U.S. debate centers on whether federal preemption is needed to prevent a patchwork of 50 state rules, what guardrails (if any) apply to the largest frontier models, how to protect children and consumers from AI-driven harms, and how to preserve American leadership against the very real possibility that China overtakes the U.S. in capability.

Where we agree
Where Most Americans Agree
  • AI is a transformative technology with both significant benefits and significant risks
  • U.S. leadership in AI matters for the economy and national security
  • Children and consumers deserve protection from clear AI-driven harms (CSAM, fraud, impersonation)
  • AI systems that make consequential decisions (lending, hiring, healthcare) should be auditable
  • Election integrity in the face of AI-generated content matters
  • Workers displaced by automation deserve support and retraining options
  • Government use of AI on citizens should be transparent and accountable

Source · Pew Research Center AI Surveys 2024-2025, KFF AI in Health

Both sides, fairly
How each side argues it.

Understanding the full debate means reading what each side actually says, not the caricature of it.

Progressive

Progressive Perspective

  • Enforceable federal safety, transparency, and bias-testing standards for high-risk AI systems are needed now—self-regulation has not worked
  • AI systems used in hiring, lending, and healthcare must be tested for and prevented from producing discriminatory outcomes
  • Workers and creators whose data trains AI deserve compensation, consent, and protection from displacement
  • Concentration of AI power in a few large corporations and a few wealthy nations is itself a public-interest problem
  • States like California and Colorado have moved ahead because Congress has not; broad federal preemption that wipes them out would be a step backward
  • Government use of AI to surveil citizens or make benefits decisions needs strong civil-liberties protections
Conservative

Conservative Perspective

  • Heavy-handed AI regulation risks ceding U.S. leadership to China, with serious national-security consequences
  • A patchwork of conflicting state AI laws makes it impossible to build national products and should be preempted by a light-touch federal framework
  • Markets, tort liability, and existing law (civil rights, fraud, product safety) cover most AI harms already
  • Mandating 'fairness' tests risks turning into ideological speech control over AI outputs
  • American AI companies should be supported, not burdened, in a global race
  • Parents and individuals, not federal agencies, should be the primary decision-makers about AI use in personal contexts
The evidence
Evidence-Based Facts
  1. 01

    In January 2025 the Trump administration rescinded the 2023 Biden AI executive order and issued a new framework emphasizing American AI leadership and reduced regulatory burden

    Source · White House executive orders; Congressional Research Service

  2. 02

    By 2025, hundreds of AI-related bills had been introduced across U.S. state legislatures; California, Colorado, Texas, and New York have enacted some of the most significant

    Source · National Conference of State Legislatures AI tracker

  3. 03

    The EU AI Act, fully effective in stages through 2026-2027, sets risk-tier obligations on AI providers including outright bans on certain uses

    Source · European Commission AI Act overview

  4. 04

    Federal agencies—including DOD, NIST, and the FTC—have issued AI guidance under existing authorities; the AI Safety Institute (NIST) coordinates voluntary frontier-model evaluations

    Source · NIST; FTC; DOD

  5. 05

    Independent surveys find roughly 6 in 10 Americans favor more government regulation of AI, with majorities in both parties concerned about misuse

    Source · Pew Research Center 2024-2025

Read deeper
Learn More
Honest questions
Questions for Thoughtful Debate
  1. 01

    Should Congress preempt state AI laws to create a single national framework?

  2. 02

    What categories of AI use (if any) warrant outright bans or pre-deployment approval?

  3. 03

    How do we keep America competitive in AI without abandoning safety guardrails?

  4. 04

    Who should be liable when an AI system causes harm—the developer, the deployer, or the user?

  5. 05

    How should AI-generated content be labeled, especially in political contexts?

  6. 06

    What protections do workers and creators need as AI reshapes their industries?

Discussion

Sign in to join the conversation

More to explore

Keep going.