Two Democratic senators are attempting to force a vote Thursday on a bill requiring artificial intelligence firms to submit their new models for government approval.
Sens. Mark Warner of Virginia and Brian Schatz of Hawaii will introduce a bill mandating that frontier companies such as Anthropic and OpenAI have the National Security Agency assess their models prior to release. Currently, the Trump administration has set up only an optional regime for testing the models.
Warner and Schatz will attempt to get the bill passed through unanimous consent, a procedure that allows for expedited votes on legislation if every senator agrees.
The push reflects the growing urgency of concerns about the pace of AI development. Sen. Chuck Grassley (R-Iowa) on Wednesday also explored using this mechanism to pass a bill to protect whistleblowers who disclose information about AI risks.
Trending
The efforts set the stage for a potential confrontation within the Senate over the risks of AI as the upper chamber prepares to adjourn for recess. The concerns about AI have been magnified in recent weeks by the disclosure of repeated hacking incidents and the high-profile resignation of a researcher at Anthropic.
Warner has been in close contact with OpenAI CEO Sam Altman, Anthropic CEO Dario Amodei and business leaders such as JPMorgan Chase CEO Jamie Dimon about the potential risks of AI and the need for federal intervention, a spokesperson said.
The Virginia senator has kept his distance from calls by his liberal colleagues such as Sen. Bernie Sanders (I-Vermont) to halt the development of artificial intelligence, calling himself “long-term bullish on AI.” But he and Schatz have pushed for more oversight than the Trump administration is supporting.
Their legislation includes other AI safety measures, such as establishing a federal database for AI safety incidents; directing the National Institute of Standards and Technology to develop an AI risk-assessment system; and authorizing the Commerce Department to work with foreign countries on testing, security and reporting.
“I’m less worried about some science-fiction scenario where AI ends humanity than I am about a very real scenario where a powerful AI system breaks into a bank, disrupts a water system, or compromises some other piece of critical infrastructure,” Warner said in a statement. “Those risks aren’t theoretical anymore, and we need to make sure our security and safeguards keep pace with the technology.
Author
