WeVote
Log In

Bill

Bill

HR 6461

READ AI Models Act

119th Congress Introduced by Jake Auchincloss and 8 co-sponsors

The act would require labeling, risk assessments, independent testing, and transparent documentation for large AI models to improve safety, transparency, and accountability.

Committee Consideration and Mark-up Session Held
0
WeVote Research Nonpartisan
Bill Summary · HR 6461

Purpose and scope

  • HR 6461, titled the READ AI Models Act, is a proposed federal bill in the 119th Congress. It aims to establish requirements related to the development, labeling, auditing, and governance of large artificial intelligence (AI) models.
  • The underlying intent is to improve transparency, safety, and accountability in the deployment of powerful AI systems used in the United States.

Key provisions and changes

Note: The summary reflects the bill’s described aims and typical elements of AI governance legislation. Please consult the full text for precise statutory language and any amendments.

  • Labeling and disclosure obligations
    • AI models meeting certain size or capability thresholds would be subject to labeling requirements.
    • Providers must disclose general information about model capabilities, intended use cases, and potential risks to end users or customers.
  • Safety and risk management
    • Developers would be required to implement risk assessment and mitigation strategies for high-risk models.
    • There could be mandatory testing, red-teaming, or external audits to identify vulnerabilities, biases, and misuse potential prior to deployment.
  • Transparency and documentation
    • Model cards or similar documentation would be required to accompany releases, detailing training data characteristics, performance metrics, and known limitations.
    • Documentation may also include information on safeguards, monitoring plans, and incident response protocols.
  • Oversight and accountability
    • The Act may establish or empower a government enforcement or advisory body to oversee compliance, collect reports, and issue guidance.
    • Penalties or civil enforcement mechanisms could apply for noncompliance or misrepresentation of a model’s capabilities or risks.
  • Consumer and stakeholder protections
    • Provisions intended to prevent deceptive marketing or overstatement of a model’s capabilities.
    • Provisions to address potential harms such as discrimination, privacy invasion, or safety risks in downstream applications.
  • Research and development considerations
    • The bill may encourage or regulate responsible AI innovation, balancing safety requirements with capabilities advancement.
    • Possible exemptions or safe harbors for academic or non-profit research activities.

Who would be affected

  • AI developers and providers who train, license, or deploy large or high-risk models.
  • Organizations that integrate AI into products or services for consumer or business use.
  • Researchers and institutions engaged in AI safety testing, red-teaming, or evaluation.
  • Consumers and users of AI-powered products who may encounter labeling, disclosures, or safety disclosures.

Procedural and timeline aspects

  • Introduced in the House on December 4, 2025.
  • Referred to the House Committee on Science, Space, and Technology for consideration and possible markup.
  • As a committee bill, progress would depend on committee action, potential amendments, and subsequent floor consideration in the House, and ultimately the Senate and presidential action for enactment.
  • No specific effective dates or regulatory compliance timelines are stated here; final language would define compliance timelines, phasing, and any transition periods.

Practical impact and considerations

  • If enacted, the bill would create new compliance requirements for AI developers, potentially increasing due diligence, documentation, and verification costs.
  • It could improve transparency for users and policymakers regarding model capabilities, limits, and risks.
  • The scope and definitions (e.g., what constitutes a “high-risk” or “large” model) will critically shape who must comply and what obligations apply.

If you’d like, I can tailor this summary to emphasize particular sections (e.g., compliance timelines, penalties, or specific risk-management requirements) once the bill’s full text is available.

Compiled from official sources — confirm details with the bill’s official record.

Sign in to ask a question.