Skip to content

AI Warnings Go Unanswered in Congress

Picture of Eric Zalcman

Congressional rhetoric on artificial intelligence (AI) has surged in recent weeks, driven by Anthropic CEO Mark Amodei’s call for frontier-wide regulation and the public warnings by Anthropic researchers on the technology’s threat to humanity. Despite other AI leaders concurring with Amodei’s prescription for national legislation, Capitol Hill has not taken the hint. With the House already gone and the Senate in its final days of session, the urgency will not translate into any legislative action before the midterms—and likely not until the next Congress.

House: A cohort of House Democrats with existing proposals for AI regulation have used this month’s developments to advance their efforts. That included a letter from 107 House Democrats demanding Speaker Mike Johnson to cancel the recess until legislation creating AI safeguards was voted on. Though their plea was denied, three bipartisan bills to govern catastrophic risk dominate discussions in the lower chamber:

  • FRONTIER Act: Reps. Lori Trahan (D-MA) and Jay Obernolte (R-CA) seek to require the largest AI developers to publish safety frameworks and undergo independent, federally licensed audits. Their measure would also give the federal government sole authority to order emergency restrictions on models posing imminent catastrophic risk, with penalties up to $10 million per violation per day.
  • AI Kill Switch Act: Reps. Ted Lieu (D-CA) and Nathaniel Moran (R-TX) propose mandating the same frontier firms to have the technical capacity to “throttle, suspend, or fully shut down” their models. The legislation would also empower the Department of Homeland Security to order developers to halt such systems.
  • Stop Rogue AI Act: Reps. Josh Gotheimer (D-NJ) and Mike Lawler (R-NY) defer to the National Institute of Standards and Technology to develop and enforce “national standards, guidelines, and best practices for discovering, verifying, and controlling AI agents.”

A subset of this group has proposed to House Minority Leader Hakeem Jeffries that Democrats should establish a Select Committee on AI if they take control of the chamber following the midterms, though the idea has already received backlash from members in committees with existing jurisdiction over the issue. Reps. Russell Fry (R-SC) and Suhas Subramanyam (D-VA) are separately standing up an “Innovators Caucus” geared toward supporting small and mid-size AI developers, an innovation-first counterweight to the current focus on regulating larger firms.

 

Senate: A working proposition by Sens. Amy Klobuchar (D-MN) and Ted Cruz (R-TX) that would hold AI firms liable to mitigating harms caused by their models has stalled due to disagreements with Sen. Maria Cantwell (D-WA), Ranking Member of the Commerce Committee. Although the bill is backed by Senate Majority Leader John Thune, Cantwell remains firm on amending aspects of the bill that would rely on AI firms themselves to conduct safety tests on their models, rather than entrusting that oversight to federal agencies.

  • Sen. Josh Hawley (R-MO) opened a formal investigation into OpenAI on the Hugging Face breach, as well as “growing allegations of the existential risk of new AI products.” He was joined by Sen. Chris Van Hollen (D-MD), who separately pressed Altman to give federal cyber agencies direct access to assess OpenAI’s model risk.
  • Following the breach, Sen. Bernie Sanders (I-VT) and House Progressive Caucus Chair Rep. Greg Casar (D-TX) announced they would be introducing the Ban Artificial Superintelligence Act that would “permanently ban the development and deployment of superintelligent AI and temporarily pause advanced AI development until a federal regulator has established safety rules.”
  • Sen. John Kennedy (R-LA) introduced the AI Emergency Button Act, requiring developers to install human-controlled mechanisms to shut down advanced models. Sen. Rand Paul (R-KY) objected to its passage by unanimous consent, arguing an “open-ended” mandate without comprehensive study would stifle the technology’s deployment. Rep. Tom Kean (R-NJ) leads the House companion of the bill.
  • Sen. Mark Kelly (D-AZ) introduced the “Make AI Work for Americans Act,” which would create a federal fund for the welfare and retraining of displaced workers through excise taxes on digital advertising, AI usage, and excess profits.
  • Sen. Jacky Rosen (D-NV) has called on Cruz, who chairs the Commerce Committee, to hold a hearing with AI executives on “the dangers posed by the pace of AI development and the guardrails needed to keep Americans safe.”

President Donald Trump and House Speaker Mike Johnson have both pushed back on government guardrails on AI and expressed preference for corporate responsibility, warning regulation would cede ground to China. Last week’s summit between Trump and Chinese President Xi Jinping avoided agreement on addressing AI threats. Trump, Johnson, and Amodei, as well as other tech CEOs, will hold a meeting on Sept. 29 at the White House to discuss AI. Without support from the administration and House leadership, there is little to no room for congressional action before the Senate leaves at the end of this week. The House returns November 9 after the midterm elections.

 

The content in this article represents the views and opinions of the individual author and does not necessarily represent the views or opinions of the Mackinder Forum.