• Unions are pushing AI into workplace bargaining, while states move to keep humans in decisions.
• Employers and tech firms warn rules could slow productivity.

AI is moving quickly into the workplace — from offices and warehouses to hospitals and public roads. Employers see a chance to boost efficiency, but unions are warning about the risks to jobs and worker rights. In statehouses across the country, lawmakers are now weighing bills that could shape how much control companies have over the technology.

The movement to regulate AI in the workplace has gained speed since the Senate killed a proposed federal moratorium earlier this year that would have blocked states from passing AI laws for the next decade. With Washington stepping back, unions and local governments are filling the gap.

How AI in the workplace sparked a video game strike

The anxiety around AI is not abstract. In July 2024, video game actors walked off the job after talks between SAG-AFTRA and companies like Activision, Electronic Arts, and Insomniac Games broke down. The dispute lasted 11 months, fueled by concerns that studios would replace human voices and performances with machine-generated replicas.

That strike ended this summer, when union members approved a new contract. The deal gives performers a 15.17% pay increase right away, followed by additional 3% raises over the next three years. More importantly, it sets limits on how companies can use AI.

Any attempt to create a digital version of an actor now requires consent. Studios must also disclose when replicas are used, and actors have the power to suspend that consent later if they feel it is being misused.

“Now that the agreement is ratified, video game performers will be able to enjoy meaningful gains and important AI protections, which we will continue to build on as uses of this technology settle and evolve,” said Duncan Crabtree-Ireland, the union’s chief negotiator.

[See also: How to Sell Your Games Overseas – in Conversation with Worldline]

The strike highlighted the stakes for performers. Earlier this year, a leaked PlayStation experiment showed AI powering responses for the character Aloy in Horizon Forbidden West. Ashly Burch, who voices Aloy, worried about what could happen if studios moved forward with such tools. “The possibility that if we lose this fight, that person would have no recourse,” she said. “They wouldn’t have any protections, any way to fight back.”

Federal rollbacks leave states to act

While unions are pressing their concerns at the bargaining table, the federal government has shifted in the opposite direction. On his first day back in office, President Donald Trump revoked several Biden-era executive orders on AI. Those orders had called for responsible AI use, risk assessments, and safeguards against discrimination in federal hiring.

Trump replaced them with his own directives. One order told agencies to base hiring decisions on skills and experience rather than factors such as race or gender. Another instructed agencies to pull back existing AI regulations and policies. Within days, the Equal Employment Opportunity Commission and the Labour Department removed their guidance on AI discrimination from their websites.

That left states to take the lead. Lawmakers in several states argue that if Washington won’t step in, they must protect workers from AI tools that could quietly reinforce bias or strip away human oversight.

• New York City began enforcing its AI Bias Law in 2023, requiring independent audits of hiring algorithms every year.
• Illinois passed its AI Video Interview Act in 2020, forcing employers to get consent before analysing job applicants through AI.
• Colorado went further in 2024, passing the nation’s first broad law on “high-risk” AI systems. Starting in 2026, employers and insurers will have to test these systems, prevent discrimination, and explain decisions to consumers.

Other states—including California, Maryland, Texas, and Virginia—have bills on the table that would require impact assessments, ban certain facial recognition tools, or create reporting duties for AI developers.

Colorado becomes a test case

Colorado’s law on AI in the workplace has drawn national attention because of its scope. It doesn’t just cover hiring—it also extends to housing, health care, financial services, and legal decisions. Supporters say it creates basic transparency for people whose lives are shaped by algorithmic choices.

But the tech industry has fought back. Companies argued that the law’s requirements would burden startups and discourage innovation. In response, lawmakers are now rewriting sections of the statute. A bill introduced this spring raises the threshold for which companies must comply—from 50 employees to 500—and adjusts how discrimination is defined. Still, critics say the revisions don’t go far enough.

“The bill substantially heightens the costs and administrative burdens on small businesses,” said Bryan Leach, CEO of Denver-based digital coupon company Ibotta. Supporters counter that the industry simply does not want to be regulated at all.

Whatever the outcome, other states are watching. If Colorado manages to enforce its law without chasing away investment, its rules could become a model nationwide.

[See also: 80% of CISOs call for regulation of DeepSeek in the UK]

Unions demand workplace AI safeguards in contracts

Beyond legislation, unions are making AI a central issue in negotiations. The AFL-CIO has launched a national task force to help unions address the technology in contracts and to draft model bills for state lawmakers.

Lorena Gonzalez, who co-chairs the task force, likens the moment to the 1990s, when NAFTA reshaped American manufacturing. “We cannot allow AI and technology to be our next NAFTA,” she said on a podcast.

The Teamsters have taken a similar stance in Massachusetts, backing a bill that would require human drivers in autonomous vehicles. In California, lawmakers are weighing a proposal that would prevent employers from using AI alone to discipline or fire workers. The measure would also restrict tools that try to predict an employee’s emotional state or behaviour.

Courts and liability risks

Even without new laws, courts are already weighing how far liability extends when AI is used in hiring. In 2024, a federal judge in California allowed a discrimination case against Workday’s recruitment software to proceed. The case could become a class action, raising the possibility that not only employers but also software developers may be held responsible for biased outcomes.

Legal experts say this shows that employers cannot hide behind the claim that an algorithm made the decision. Human oversight remains crucial, and ignoring it could put companies at risk.

AI in the workplace: Productivity gains or job losses?

Not everyone agrees that more rules are the answer. Some business advocates argue that automation has always displaced certain jobs while creating new ones. Rob Atkinson, president of the Information Technology and Innovation Foundation, said that restricting AI could slow wage and economic growth. “Rather than engaging in a Luddite-inspired campaign against a critical new technology, organised labour should push Congress to improve programs for displaced workers,” he wrote in a statement.

Still, polls suggest many workers are uneasy. A Gallup survey earlier this year found that one-third of American workers fear AI will reduce job opportunities. For unions, that fear is enough reason to demand protections now rather than wait to see how the technology unfolds.

A shifting balance

The clash over AI has made its way into statehouses, union halls, and courtrooms, and it shows no signs of slowing. With federal policy rolling back and state laws moving ahead, businesses face a patchwork of rules and rising pressure from workers.

As Massachusetts State Sen. Paul Feeney put it while defending his autonomous vehicle bill: “We can’t just be the Wild West where tech companies can experiment on our roads unregulated. Let’s make sure we’re introducing commonsense regulations that keep people safe and employed.”