The discussion draft, obtained by The Hill, proposes overriding state regulations that target AI model development for three years. This would not necessarily preempt state laws dictating how AI is used once released, according to the text.
The draft “expressly does not preempt laws of general applicability, common law remedies, or laws regulating AI use or deployment,” a summary of the draft stated.
The prospect of an AI preemption law has been in talks for more than a year, but Washington has largely stalled on getting anything across the finish line. The Senate came close to passing a moratorium on new state AI regulations last year, but it fell through at the last minute.
The proposal would also codify the Center of AI Standards and Innovation (CAISI), which was launched last June under the Department of Commerce.
CAISI would oversee voluntary guidelines and standards for AI security, along with licensing independent verification organizations to audit compliance from labs.
Larger frontier developers would also face new transparency requirements to publish a “frontier AI framework” including the technical and organization protocols it uses to evaluate and manage a model’s “catastrophic risks.”
The labs would also be required to file a report with CAISI shortly after certain safety incidents, including those posing “imminent risk of death or serious injury.”
The pair of lawmakers also published an op-ed in Bloomberg Law on Thursday morning, emphasizing the discussion draft is not a “final product,” but rather a way to start a broader conversation.
Read more in a full report at TheHill.com.
