RISE Act of 2025

Summary#

The RISE Act of 2025 would create conditional civil immunity for certain artificial intelligence developers when their AI product makes an error while being used by a licensed, learned professional to provide professional services. To qualify, a developer must publicly release and keep updated a model card and a model specification for the AI product, and give clear documentation to professionals about known limits, failure modes, and appropriate uses. Developers may redact trade secrets unrelated to safety, but must provide a written justification for each redaction. Immunity does not cover recklessness, willful misconduct, fraud, knowing misrepresentation, or uses outside the scope of professional services. If a developer fails to update disclosures within 30 days after deploying a new version or learning of a new material failure mode, and that failure to update is a proximate cause of harm, immunity does not apply. The bill would preempt state law claims covered by this immunity. The act would take effect on December 1, 2025, and apply to acts or omissions on or after that date.

What it means for you#

• If you are an AI developer: You could be protected from some civil lawsuits for AI errors when a licensed professional uses your product, but only if you publish and maintain the required technical documents, explain limits to professionals, avoid reckless or willful misconduct, and update disclosures on time. You may redact certain trade secrets but must justify those redactions in writing.
• If you are a learned professional (for example, a licensed doctor, lawyer, or other credentialed expert): You must rely on your professional judgment when using AI tools. The developer must provide model cards, specifications, and documentation about limits and failure modes that you can use to assess safe use.
• If you are a client of a learned professional: The bill affects who can be sued for harm caused by AI errors used in professional services. Claims based on fraud or knowing misrepresentation would still be allowed.

Expenses#

No publicly available information.

Proponents' View#

The bill's findings say clearer standards for AI transparency, together with a defined liability framework, will reduce uncertainty, help integrate AI into professional services, and balance innovation with safety. Supporters view public model cards and specifications as a way for professionals to understand AI capabilities and limits.

Opponents' View#

No publicly available information.