Skip to content
News

ARI Endorses Bold New AI Regulatory Framework Proposal from Bipartisan Trio

On Thursday, Reps. Sara Jacobs (D-CA), Celeste Maloy (R-UT), and Don Beyer (D-VA) released details on a bold, new bipartisan AI regulatory framework to mitigate the risks posed by rapidly advancing AI systems. The proposed Safe and Accountable Federal Evaluations of Artificial Intelligence (SAFE AI) Act would include mandatory AI safety standards, government-backed assurance to ensure developers meet those standards, transparency into model development and training, incident reporting, and strong enforcement mechanisms.

Reps. Jacobs, Maloy, Beyer propose SAFE AI Act

On Thursday, Reps. Sara Jacobs (D-CA), Celeste Maloy (R-UT), and Don Beyer (D-VA) released details on a bold, new bipartisan AI regulatory framework to mitigate the risks posed by rapidly advancing AI systems. The proposed Safe and Accountable Federal Evaluations of Artificial Intelligence (SAFE AI) Act would include mandatory AI safety standards, government-backed assurance to ensure developers meet those standards, transparency into model development and training, incident reporting, and strong enforcement mechanisms.

The proposal offers preemption as a floor rather than a ceiling for state laws, allowing states to adopt more protective requirements and to regulate AI models and developers the bill does not reach. Americans for Responsible Innovation (ARI) endorsed the proposed framework, which reflects the principles in ARI’s Blueprint for AI Governance.

“It’s time to close the book on the era of voluntary company commitments and put hard-and-fast rules in place that demand safety, transparency, and accountability from our leading AI companies,” said ARI President Brad Carson. “This is a gold standard proposal with bipartisan support that should guide the work of the next Congress as lawmakers debate how to regulate AI. The AI incidents we’ve seen over the last year and voter demands for tech accountability in the midterm elections have given Congress a mandate to act. This is exactly the kind of serious, substantive, bipartisan effort that meets the moment.”

The SAFE AI Act includes a range of safety, transparency, and accountability measures, including:

  • Government-set standards: The Center for AI Standards and Innovation (CAISI) would set minimum standard requirements for the safety frameworks of AI developers.
  • Enforceability: Unlike voluntary company commitments, approved standards under the bill would be legally binding and could not be unilaterally loosened.
  • Pre-deployment model evals: Federal evaluation teams, or accredited IVOs, would evaluate models to ensure framework compliance before commercial release or broad internal access.
  • Assurance: Federally accredited independent verification organizations (IVOs), would review frameworks, conducting compliance evaluations with the federal government having final authority. Evaluations would happen before commercial release of models.
  • Incident reporting: Developers would be required to report safety incidents, including imminent risk of severe harm.
  • Transparency on AI R&D: Developers must disclose substantially autonomous AI R&D using prescribed measures.
  • Whistleblower protections: Whistleblowers would be protected with the ability to report incidents to CAISI.
  • Emergency authorities: The government could order mitigation when safeguards are inadequate and a system poses an imminent risk of severe harm, subject to prompt judicial review.
  • Preemption: States can regulate above the federal floor.

In August, ARI released its own proposed blueprint for AI governance, a federal policy framework for addressing the growing risks from frontier AI. The blueprint is built around three governance functions any federal proposal should incorporate: government-set safety standards, government-backed assurance, and transparency into frontier model development and the automation of AI research.

###