Congress stalls on AI rules as leading proposals focus on catastrophic risks

The AI bills getting attention in Washington this week are about nuclear and biological weapons, models that escape human control, and whether the federal government can block a release. None of the leading proposals described in the week's reporting creates a national rulebook for employer uses of AI such as hiring, monitoring or displacement; the closest labor-related measure cited is Hawley's federal-tracking bill tied to labor protections.

That gap sits inside a fight that is going nowhere quickly. President Donald Trump spent Monday dismissing the case for guardrails, calling fears about the technology a "HOAX" and writing that "the only control or 'guardrails' that AI needs is a STRONG AND SMART (High IQ!) PRESIDENT, and the U.S.A. has that, in spades!", according to CNBC's account of his social media posts. He has separately described efforts to limit the technology as a "conspiracy."

The calls he was answering came from the industry itself. Anthropic's Dario Amodei published an essay over the weekend urging AI companies to slow the pace of frontier development and proposing a plan for auditing and regulating the technology without, in his words, "sacrificing commercial advantage" or harming U.S. national security. OpenAI's Sam Altman and xAI's Elon Musk took the unusual step of agreeing publicly, the Associated Press and CNBC reported. Rahm Emanuel, chief of staff to Barack Obama during the financial crisis, told the AP: "In my entire time of public service, I've never had an industry or a CEO ask for regulation and oversight."

The arithmetic in Congress is unforgiving. Any bill needs both chambers and, most likely, the president's signature, and Republicans hold majorities in both, the BBC noted. The House is scheduled to leave at the end of this week and not return until after the November 3 midterms; the Senate is expected to have three voting weeks, according to Reuters. House Speaker Mike Johnson, asked whether the chamber would act before the break, said only: "This is a very complex issue."

What is being negotiated matters more to workforce leaders than whether it passes. Senate talks involving Majority Leader John Thune, Commerce Chairman Ted Cruz and Senator Amy Klobuchar would create a "duty of care" requiring developers to design against "catastrophic risks," and would give the government power to block release of certain models, subject to challenge in federal court, Reuters reported, citing Senate aides and a lobbyist involved. Part of the measure would also block states from enforcing their own laws covering certain model risks. Cruz has said publicly the work is aimed at "catastrophic risks involving biological or nuclear threats."

State law is where AI rules currently bite. Regulation of the technology in the U.S. happens mostly at the state level, CNBC reported, citing California's recent package of AI bills as an example. The scope of any federal preemption is still being negotiated, which makes the question of which state authorities survive a live one for multistate employers and their vendors.

The closest thing to a labor provision in the federal debate comes from Senator Josh Hawley, who has introduced bipartisan legislation requiring federal tracking of AI safety concerns tied to national security, civil liberties and labor protections. "We need to protect working people," Hawley said. In the House, the FRONTIER Act from Representatives Jay Obernolte and Lori Trahan would require independent audits of AI research labs and let the government halt use of a model where officials find an "imminent catastrophic risk."

Adjustment policy has fared no better than safety policy. A bipartisan working group formed by then-Senate Majority Leader Chuck Schumer recommended in 2024 that the U.S. spend at least $32 billion over three years developing AI and building safeguards around it; there was little follow-up, the AP reported. Congress has neither imposed rules nor funded ways for people to adjust to the potential changes from AI.

The executive branch has moved in the other direction. Trump rescinded the Biden administration's 2023 executive order on AI oversight soon after returning to office. In June he signed an order asking AI companies to voluntarily submit models to the government before release, CNBC reported. Alexandra Reeve Givens, who leads the Center for Democracy and Technology, told the BBC that the voluntary framework itself remains a secret and that requests to release it have gone unanswered: "There's no clear standard to make sure that the approach is even grounded in the rule of law."

For anyone running hiring or workforce systems in the meantime, the most consequential observation in the week's reporting came from Timnit Gebru, founder of the Distributed AI Research Institute, who argued that the existing statute book already reaches most of what AI does commercially. "There is no AI exemption in existing statutes," she told CNBC. "Fraud is still fraud. Discrimination is still discrimination."