RISE Act of 2025

Bill Number

S. 2081

Origin Chamber

Senate

Congress

119th Congress, Session 1

Policy Area

Science, Technology, Communications

Status

Introduced

Latest Action

2025-06-12: Read twice and referred to the Committee on Commerce, Science, and Transportation.

Last Updated

2025-07-08T13:21:18Z

AI-Generated Summary

Purpose of the Legislation

The Responsible Innovation and Safe Expertise Act of 2025 (RISE Act) aims to encourage the safe and responsible use of artificial intelligence (AI) in professional services like healthcare, law, and finance. It does this by providing limited legal protection (immunity) to AI developers from lawsuits, but only if they share clear information about their AI products' strengths, weaknesses, and risks. This is intended to reduce uncertainty that slows AI adoption while ensuring professionals and clients can better understand and manage AI risks.

Key Provisions

• Definitions: The Act defines key terms to clarify its scope:
• Artificial intelligence: Refers to systems that perform tasks like perceiving, learning, or decision-making, as defined in existing federal law (National Artificial Intelligence Initiative Act of 2020).
• Developer: Any person or company that creates, modifies, controls, or markets an AI product.
• Learned professional: Licensed experts (e.g., doctors, lawyers) bound by ethical duties and professional standards when serving clients.
• Client: Someone who hires a learned professional and relies on their advice under professional rules.
• Error: AI outputs or failures that are false, misleading, or harmful in ways a reasonable developer could predict, or failures to do what the AI claims it can.
• Model card: A public document detailing the AI's training data, performance tests, intended uses, limits, and safeguards against errors.
• Model specification: Instructions or prompts that guide the AI's basic behavior, including hidden system prompts or training guidelines not seen by users.

• Conditional Immunity for Developers (Section 4): AI developers are protected from civil lawsuits (e.g., for damages caused by AI errors) when their product is used by a learned professional for client services, but only if:
• They publicly release and keep updated a model card and model specification before deployment (redactions allowed only for true trade secrets, with written explanations).
• They provide clear documentation to professionals about the AI's known limits, failure points (e.g., when it might give wrong advice), and suitable uses.
• The immunity applies only to non-reckless or non-intentional misconduct by the developer.
• Developers must update these materials within 30 days of releasing a new AI version or discovering a major new flaw; failure to do so ends immunity if it causes harm.

• Preemption of State Laws (Section 4(d)): Federal immunity overrides conflicting state laws for covered claims, but does not block state lawsuits for fraud, deliberate lies, or AI use outside professional settings.

• Preservation of Other Protections (Section 5): The Act does not change existing federal, state, or common law immunities unrelated to this AI-specific protection.

• Effective Date (Section 6): Takes effect December 1, 2025, and applies only to actions or errors after that date.

Significant Changes to Existing Law

• This is the first federal law to offer targeted immunity specifically for AI developers in professional contexts, shifting some responsibility from creators to licensed users (learned professionals).
• It introduces mandatory transparency requirements (e.g., model cards and specifications) as a condition for protection, which were not federally required before.
• It preempts state liability rules for qualifying cases, creating a uniform national standard where none existed, but carves out exceptions for misconduct like fraud to align with current tort (civil wrongdoing) laws.
• No changes to criminal liability or non-professional AI uses; it builds on but does not alter the 2020 AI Initiative Act's definition of AI.

Potential Impacts

• On Government Agencies: Federal agencies (e.g., those overseeing AI under the Commerce Department) may need to monitor compliance and develop guidelines for "model cards," potentially increasing administrative workload. It could influence broader AI regulation by favoring transparency over strict bans.
• On Citizens/Clients: Clients of professionals (e.g., patients or legal clients) may benefit from safer AI integration due to better disclosures, but they remain protected by professionals' existing duties of care. Everyday users outside professional settings see no direct change.
• On AI Developers and Businesses: Encourages innovation by reducing lawsuit fears, but requires upfront investment in transparency; non-compliance could lead to more litigation.
• On Learned Professionals: Provides clearer tools to evaluate AI, helping them meet ethical obligations, but increases their potential liability as primary decision-makers.
• On International Relations: Minimal direct impact, as it's domestic-focused, but U.S. transparency standards could set a global example or influence trade in AI tech, potentially pressuring foreign developers to adopt similar practices for U.S. markets.

Main Stakeholders Affected

• AI Developers: Primary beneficiaries of immunity, but must invest in disclosures; includes tech companies creating or selling AI tools.
• Learned Professionals: Doctors, lawyers, accountants, etc., who use AI; they gain better information but bear more responsibility for errors.
• Clients: Individuals or businesses relying on professional services; indirectly protected through professional standards.
• State Governments and Courts: Affected by preemption, which limits state-level lawsuits and may reduce state court caseloads for AI claims.
• Industry and Regulators: Tech firms, professional associations (e.g., American Medical Association), and federal bodies like the FTC or NIST, which may enforce or guide transparency.

Notable Legal, Constitutional, or Political Implications

• Legal Implications: Shifts liability framework from strict developer accountability to a "safe harbor" based on transparency, potentially reducing frivolous lawsuits but raising questions about whether disclosures are sufficient to prevent harm. It preserves claims for serious misconduct, maintaining balance in tort law.
• Constitutional Implications: The preemption clause invokes federal supremacy over states (under the Supremacy Clause of the U.S. Constitution), which is routine for national tech policies but could face challenges if seen as overreaching into state professional regulations. No direct free speech or privacy issues, as disclosures are factual and optional for immunity.
• Political Implications: Promotes a pro-innovation approach by limiting liability, aligning with bipartisan efforts to regulate AI without stifling growth (introduced by Sen. Cynthia Lummis, R-WY). It emphasizes voluntary transparency over mandates, which could spark debate on whether it adequately protects vulnerable clients or favors big tech. Referred to the Senate Commerce Committee, it signals growing congressional focus on AI governance amid rapid tech advances.

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

Sponsor

Sen. Lummis, Cynthia M. [R-WY]

Recent Actions

• 2025-06-12: Read twice and referred to the Committee on Commerce, Science, and Transportation.
• 2025-06-12: Introduced in Senate

Bill Versions

• Responsible Innovation and Safe Expertise Act of 2025 — issued 2025-06-12 — PDF (10 pages)