A bill is going around Congress right now that looks like it would be easy to understand. For businesses that have to lay off workers because of AI, they need to tell the Department of Labor. Four times a year. Data for everyone. A trail of paper. The Effects of AI on Jobs The Clarity Act, which was introduced by Congressman Steven Horsford of Nevada, Congresswoman Sara Jacobs of California, and Congressman James Moylan of Guam, is a bill that would require big businesses and the government to say when AI takes over jobs that used to be done by people. It’s the kind of idea that seems obvious in 2026, but it still feels like it’s coming too late.
The bill builds on work done by Senators Mark Warner and Josh Hawley in November 2025, when they introduced similar legislation. In their version, they asked publicly traded companies, some private companies, and federal agencies to report every three months on job losses, new hires, fewer hires, and efforts to retrain workers caused by AI. This would be put together into a public report by the Department of Labor. Warner said it would help them get “a clear picture.” Hawley talked about predictions that AI could make unemployment reach 20% in five years. Neither of them sounded casual about it.
Now is different, and it’s not just the text of the law. It has to do with the mood. Earlier this year, Dario Amodei, CEO of Anthropic, said that AI tools could eliminate half of all entry-level white-collar jobs. In one month, Amazon, UPS, and Target all said they were letting go of more than 60,000 jobs. Some of those layoffs were linked to AI, while others were linked to tariffs or changes in business priorities. There isn’t much difference anymore between saying “AI took your job” and saying “we restructured, and AI was a part of it.” Companies can smoothly cross that line. That’s the exact hole that this law aims to fill.
But big business in America isn’t going to sit still. Along the same lines, OpenAI, Anthropic, Amazon, Microsoft, and others have helped a separate effort led by Gina Raimondo, former Secretary of Commerce, raise $500 million for a nonprofit called Raise Us. The group says it will help workers get used to AI by offering retraining programs, wage insurance for workers who lose their jobs, and partnerships with community colleges across the state. For high school graduates, Maryland is making the “service year” longer. Microsoft is teaching new lawyers how to do other jobs so that they can be repositioned as AI changes the way lawyers do their jobs. On the surface, it looks like a nice package.
Still, I get the sense that this isn’t just charity. It’s the placement. Companies that pay for transition programs on their own gain credibility and power when they try to argue that reporting requirements that are required by law are not needed or even harmful. Raise Us mostly works with governors, not Congress. It’s not a mistake. It is easier to make partnerships between states. It’s harder to get out of federal rules. There isn’t a “exit clause” in any bill, but there are private-sector alternatives that could give lawmakers a reason to loosen the rules on reporting or put off enforcing them.
It’s still not clear if the AI-Related Job Impacts Clarity Act has enough votes. Being backed by both Democrats and Republicans is helpful, but the bill faces the same problems that have stopped almost all AI regulations in Washington. With its attention on the benefits of AI for investments and the stock market, the White House has been careful not to push too hard on worker protections. Glassdoor, a site that rates employers, has found that how workers feel about AI is getting worse over time. People are aware of what’s going on. They just can’t prove it yet because they don’t have the data. That’s exactly what this bill is meant to fix.
It’s been Senator Bernie Sanders who has talked about it more. Along with a call for a “robot tax” on companies that aggressively automate, his report warned that AI could take away up to 97 million jobs in the next ten years. The Horsford-Jacobs-Moylan bill calls for measured transparency, which is not the same thing. But the difference between those two points of view shows where this debate is going. If voluntary disclosures and retraining programs paid for by corporations don’t work, politicians will only want stricter rules.
As you walk by the lobbying firms on K Street, it’s hard not to notice that the same companies that say they will help workers understand AI are also spending a lot of money to make sure the rules about AI stay flexible. The real story here is the tension between openness and control, between telling the truth and keeping things secret. Congress needs to see the data. Businesses want to be able to decide what the data means. Millions of workers are waiting to see which side wins.
