Congress’s difficulty regulating artificial intelligence is not a shortage of interest. Lawmakers have proposed a wide range of AI measures, and a bipartisan House task force produced recommendations in 2024. The harder task is turning concern into agreement about what government should regulate—and how.

The stalemate reflects several overlapping disputes: whether new rules would protect Americans or weaken U.S. competitiveness, which parts of government should oversee AI, and what states should do while Washington debates.

The central divide: safeguards or a competitive disadvantage?

Supporters of AI regulation argue that a technology capable of influencing employment, financial decisions, public information and national security needs clear accountability. From that perspective, waiting allows deployment to outpace protections.

Opponents of expansive regulation emphasize a different risk: compliance costs and restrictions could slow American developers while competitors, especially China, move ahead. The Trump administration’s resistance to congressional regulation adds another obstacle to building a federal legislative coalition.

These positions frame the same issue differently. One treats inadequate oversight as a threat to the public; the other treats excessive oversight as a threat to innovation and national security. Agreement that AI matters does not resolve that underlying conflict.

AI does not fit neatly into one committee

Congress is organized around policy areas. AI cuts across them.

• Financial services: AI can influence lending, fraud detection and other consequential decisions.
• Energy: The infrastructure supporting AI raises questions about electricity demand and capacity.
• National security: AI capabilities have strategic and defense implications.
• Commerce: AI products and services operate across state borders and industries.

A broad AI bill can therefore involve multiple committees, agencies and competing priorities. Lawmakers must negotiate not only the substance of safeguards but also who writes the rules, who enforces them and which activities fall within their scope.

This creates a practical dilemma. Comprehensive legislation may address interconnected risks, but it requires a larger agreement. Narrow bills can be easier to negotiate, yet leave important questions unresolved.

State action creates a second policy fight

As federal lawmakers struggle to agree, states have pursued their own AI rules. That gives state governments a way to respond to emerging concerns without waiting for a national framework.

It also raises the question of how state and federal authority should fit together. Businesses operating nationwide may face different obligations across jurisdictions. Meanwhile, replacing state protections with a federal approach could become contentious if that approach offers fewer safeguards.

The key issue is not simply whether Washington should act. It is whether federal standards should supplement state rules, establish a minimum level of protection or replace some state requirements altogether.

Bipartisan interest is only a starting point

The 2024 task force shows that Republicans and Democrats can collaborate on identifying AI’s opportunities and risks. But a report is not an enforceable law.

Legislation requires agreement on definitions, obligations, exceptions and enforcement. It also needs committee action, leadership support and time on the congressional calendar. Election-year pressures can make an already complicated negotiation harder to finish.

Congress’s long-running struggles over online child safety offer a cautionary parallel: widespread concern about technology does not automatically produce consensus on a legislative remedy.

What meaningful progress would look like

Bill counts and public statements reveal interest, but they are weak measures of whether Congress is close to acting. More useful signals include bipartisan agreement on specific requirements, movement through relevant committees and a workable division of responsibility between federal and state authorities.

It is also important to distinguish the absence of a comprehensive AI framework from the absence of all legal constraints. Existing laws can still apply to conduct involving AI, depending on the activity and jurisdiction.

The central challenge is converting broad principles into workable obligations. Until lawmakers agree on what needs protection, how to preserve competition and who should enforce the rules, AI legislation will remain easier to propose than to pass.

This article was inspired by Why Congress has been stalling on AI from CNN. Please visit the original video for the creator’s full presentation and context.