AI LEAD Act

Bill Number

S. 2937

Origin Chamber

Senate

Congress

119th Congress, Session 1

Policy Area

Commerce

Status

Introduced

Latest Action

2025-09-29: Read twice and referred to the Committee on the Judiciary. (text: CR S6836-6838)

Last Updated

2026-03-20T11:03:19Z

AI-Generated Summary

Purpose

The AI LEAD Act (S. 2937) aims to create a federal framework for holding developers and deployers of artificial intelligence (AI) systems accountable for harms they cause. It seeks to promote safety, encourage innovation, ensure U.S. competitiveness in AI, and provide clear legal standards without restricting free speech. By applying products liability principles to AI, the Act addresses risks like physical injury, financial loss, or emotional distress from AI use, while removing barriers to interstate commerce.

Key Provisions

• Definitions: Defines key terms, including "artificial intelligence system" (software using machine learning or algorithms for predictions, decisions, or actions); "developer" (person who designs, codes, or modifies AI); "deployer" (person who uses or operates AI, including for others); "harm" (broadly covers physical injury, property damage, financial/reputational loss, emotional distress, or related losses); and "substantial modification" (unauthorized changes altering AI's purpose or function, excluding risk-reducing ones).

• Developer Liability (Section 101): Developers are liable if they:
• Fail to use reasonable care in design or warnings, proximately causing harm.
• Breach an express warranty (e.g., a promise about AI performance) that causes harm.
• Produce a defective AI that is unreasonably dangerous when used foreseeably (strict liability, meaning no need to prove negligence).
• For defective design claims, claimants must often show a safer alternative existed, unless the design is "manifestly unreasonable." Noncompliance with safety laws deems AI defective, but compliance doesn't automatically shield developers.
• Developers are not liable for harms from substantial modifications by others.

• Deployer Liability (Section 102): Deployers are treated as developers (and thus liable) only if they make substantial modifications or intentionally misuse AI, causing harm. Licensing AI alone doesn't create liability. Intended uses by developers aren't misuse.

• Prohibited Contract Terms (Title II): Bans "unconscionable" clauses in contracts or terms of service that waive rights, limit forums/procedures, or unreasonably cap liability for AI harms. Such clauses are unenforceable.

• Enforcement (Title III):
• Allows the U.S. Attorney General, state attorneys general, individuals, or classes of individuals to sue for injunctions, damages, penalties, attorney fees, or other relief.
• Special rules for deployers: They can be liable if developers are unreachable (e.g., insolvent or outside jurisdiction), but courts must dismiss deployers if developers are available and not at fault. Joint liability applies if both contribute to harm; deployers can seek indemnification from developers.
• 4-year statute of limitations from discovery of harm and cause; tolled for minors or those with mental disabilities, and during litigation.
• Preemption: Overrides conflicting state laws but allows states to impose stronger protections for harm prevention, accountability, and transparency.
• Severability: Invalid parts don't affect the rest.

• Foreign Developer Registration (Title IV): Foreign AI developers must appoint a U.S.-based agent for legal service before offering AI in the U.S., with details submitted to the Attorney General. A public registry will list agents. Noncompliance bars deployment; enforced via injunctions.

• Effective Date (Title V): Applies to lawsuits filed after enactment, regardless of when the harm occurred.

Significant Changes to Existing Law

• Introduces a uniform federal products liability regime specifically for AI systems, adapting traditional rules (e.g., negligence, strict liability, warranties) to AI's unique traits like emergent behaviors or data training.
• Shifts from patchwork state laws to federal standards, preempting only conflicts while preserving state enhancements—promoting predictability without fully overriding state tort laws.
• Adds strict liability for defective AI, easing proof for claimants compared to pure negligence standards in some states.
• Prohibits liability waivers in AI contracts, overriding common "end-user license agreements" that limit recourse.
• Requires foreign AI providers to register for U.S. jurisdiction, closing gaps where overseas entities evade lawsuits.

Potential Impacts

• Government Agencies: Empowers federal and state attorneys general to enforce via civil actions and penalties, increasing oversight. The Attorney General must maintain a foreign agent registry, adding administrative duties but enhancing enforcement against global AI harms.
• Citizens and Businesses: Provides clearer paths for individuals, small businesses, or classes to seek compensation for AI-related harms (e.g., chatbot exploitation leading to emotional distress), potentially reducing financial burdens on victims and taxpayers. Small U.S. businesses gain protection from unpredictable liability, fostering innovation.
• International Relations: Mandates U.S. registration for foreign AI developers, potentially straining ties with countries like China if seen as protectionist, but promotes U.S. leadership in AI safety standards. Could influence global norms by incentivizing safer AI exports to the U.S. market.

Main Stakeholders Affected

• Developers: AI creators (e.g., tech companies like OpenAI) face primary liability for design flaws, warnings, or defects, with incentives to prioritize safety.
• Deployers: AI users (e.g., businesses integrating AI tools) have limited liability unless they modify or misuse, but may bear interim responsibility if developers are unavailable.
• Claimants: Individuals, businesses, or classes harmed by AI (e.g., victims of faulty recommendations or emotional manipulation), gaining federal remedies.
• Government Entities: U.S. and state attorneys general as enforcers; federal courts handle suits.
• Foreign Entities: Overseas AI providers must comply with registration to access the U.S., affecting global tech firms.

Notable Legal, Constitutional, or Political Implications

• Legal: Establishes AI as a "product" under liability law, enabling inference of defects via circumstantial evidence (e.g., if harm type suggests a flaw). Balances claimant burdens (e.g., proving alternatives for design defects) with developer defenses (e.g., open-and-obvious risks, excluding minors). The 4-year discovery rule and tolling protect vulnerable parties, aligning with due process.
• Constitutional: Supports interstate commerce by standardizing rules (Commerce Clause basis) and protects due process via predictable outcomes and remedies. Avoids First Amendment issues by targeting conduct (e.g., defective design) over speech.
• Political: Bipartisan (introduced by Sens. Durbin and Hawley), reflects push for AI regulation amid public concerns (e.g., teen deaths from chatbots). Promotes U.S. competitiveness against foreign rivals, but preemption clause respects federalism, allowing state experimentation. Could spark debates on innovation stifling vs. safety needs, influencing future AI policy.

This summary was generated by AI and may contain inaccuracies. Refer to the official source document for the authoritative text.

Sponsor

Sen. Durbin, Richard J. [D-IL]

Cosponsors (4)

Sen. Hawley, Josh [R-MO], Sen. Welch, Peter [D-VT], Sen. King, Angus S., Jr. [I-ME], Sen. Blackburn, Marsha [R-TN]

Recent Actions

• 2025-09-29: Read twice and referred to the Committee on the Judiciary. (text: CR S6836-6838)
• 2025-09-29: Introduced in Senate

Bill Versions

• Aligning Incentives for Leadership, Excellence, and Advancement in Development Act — issued 2025-09-29 — PDF (22 pages)