If your organization is navigating AI regulation using a patchwork of state laws, you’re not alone in feeling the strain. Congress heard testimony on Wednesday that those very laws may be broken.
The House Education & Workforce Committee’s Health, Employment, Labor and Pensions Subcommittee is contemplating whether new federal laws are needed to regulate artificial intelligence in the workplace.
The hearing, “Building an AI-Ready America: Adopting AI at Work,” revealed divisions between witnesses who argued existing employment laws are sufficient and those who said workers need stronger protections against algorithmic management and surveillance.
Here are five key takeaways for HR leaders from the hearing.
State AI laws are struggling in practice
Brad Kelley, a shareholder at Littler and former chief counsel to the EEOC commissioner, told lawmakers that rushed state legislation is creating an “increasingly unworkable regulatory environment” for employers.
He pointed to Colorado’s AI Act as a cautionary tale, stating that Gov. Jared Polis asked legislators to fix the law’s flaws on the same day he signed it in May 2024. Days later, Polis, the state Senate majority leader and attorney general, sent a letter to the business community pledging to revise it. The law, originally scheduled to take effect in early 2025, has been delayed until June 2026 and may face further postponement.
New York City’s AI law, which took effect in July 2023, has been, according to Kelley, “widely criticized for being ineffective.”
In fact, the New York City Department of Consumer and Worker Protection ran an audit covering the period from July 2023 through June 2025, and found that only two automated employment decision tool (AEDT) complaints were filed during the audit’s two-year scope.
He argued these examples show what happens “when legislation gets ahead of itself” and warned Congress against similar missteps at the federal level.
Read more | Clarity for HR: Ban on state AI regulation nixed for now
Existing laws may be sufficient to address AI harms
Kelley made the case that the U.S. already has a “well-established technology-neutral legal framework fully capable of addressing most AI-related misconduct,” attributing laws like Title VII of the Civil Rights Act (1964), the Fair Labor Standards Act (1938) and the National Labor Relations Act (1935). “These laws were passed well before technology in the workplace, especially artificial intelligence—the internet as well. But these laws have been able to survive,” he said.
He told the committee that in his years at the EEOC and Department of Labor, as well as in private practice, he has never encountered employers using AI to interfere with union activity or suppress organizing efforts. “Many of the more extreme hypotheticals, like using AI to suppress union activity, remain almost entirely theoretical,” he testified, adding that most employers are “focused on using AI to improve efficiency and compliance.”
Union considerations
U.S. Rep. Summer Lee (D-Pa.) pushed back on this framing, pointing to reporting that Whole Foods, owned by Amazon, used an AI‑driven “heat map” to identify stores at risk of union activity.
She also stated that the National Eating Disorders Association informed helpline staff they would be laid off and replaced by an AI chatbot just weeks after those workers voted to unionize.
CBS News reported, however, that the bot was shut down “after it provided harmful advice to people seeking help.”
Read more: What a suit against Amazon says about RTO and HR tech
Federal agencies lack data on how AI is reshaping work
Revana Sharfuddin, a labor economist and research fellow at George Mason University’s Mercatus Center, told lawmakers that federal statistical agencies are flying blind when it comes to measuring AI’s real impact on American workers. The problem: Current labor statistics count jobs rather than measuring how work is done.
“AI operates differently” from previous waves of automation, Sharfuddin explained. “It automates specific tasks while leaving others untouched.”
She used her own work as an example. As an empirical economist, she previously spent a significant amount of time writing computer code for economic modeling. “AI now handles much of that, freeing me to focus on research design, collaboration and communication,” she said. “The occupation ‘economist’ still exists, but the work has changed. This is precisely what we need to measure.”
Three ‘achievable steps’
She cited a Stanford study finding that across 950 occupations, although AI could perform certain tasks in nearly all jobs, there was no occupation that AI could fully handle alone. Yet, the U.S. Bureau of Labor Statistics/U.S. Census Current Population Survey includes no questions about AI use or task changes.
Sharfuddin recommended three “achievable steps.” These are adding an AI supplement to federal surveys to capture how workers use AI and whether it has changed their tasks, linking firm-level adoption data to worker outcomes through existing Census programs and coordinating annual reports across federal agencies. The Government Accountability Office flagged this data gap in 2019, but “the pace of AI adoption has accelerated dramatically” since then, she said.
“Good policy requires good data, and we currently don’t have it,” said Sharfuddin, in closing.
Read more: What a suit against Eightfold says about ongoing AI in hiring risks
Algorithmic management is already pervasive
Tanya Goldman, a fellow at The Workshop, an organization that leverages policy and partnerships to protect and expand workers’ rights, and a former senior official at the Department of Labor and EEOC during the Biden administration, presented a different picture of AI adoption than Kelley.
She cited an Organisation for Economic Co-operation and Development (OECD) survey showing 90% of U.S. employers use algorithmic management tools and that two-thirds of workers report such tools are used in their workplaces “across earnings in different jobs.”
Goldman described how AI enables more invasive surveillance than any human manager could conduct. Efforts such as tracking bathroom breaks, using facial recognition to assess attention to computer screens, monitoring mouse movements and keystrokes, and controlling scheduling and task assignments were all mentioned as employee surveillance tactics during the hearing.
“Time-on-task measures can pressure workers to work constantly and make it difficult to take necessary breaks, which can penalize pregnant or disabled workers if they take a break just to use the restroom or to take some medicine,” she testified.
AI for employment-related decisions
In hiring, Goldman said, one-third to “the vast majority” of employers use AI tools for recruitment and screening. She said these systems often lack transparency, meaning “workers often do not understand how decisions affecting their livelihoods are being made. This can leave them without meaningful recourse to challenge the decision or to correct inaccurate data that is being used.”
She pointed to algorithmic wage-setting that can suppress pay, citing a Roosevelt Institute report on nurses: “On the same day at the same hour in the same hospital, two different gig nurses can be paid different amounts by the same app.” The tool referenced here is ShiftKey, which helps gig nurses find and accept hospital shifts. Roosevelt Institute referred to it as “Uber for nursing.”
Goldman argued these harms are “the product of choices, not an inevitability” and called for federal guardrails, including transparency requirements, mandatory human involvement in significant employment decisions, bias testing and a meaningful role for workers and unions in AI implementation.
Read more: DOL funding largely survives proposed cuts in 2026 budget
Worker protection agencies lack enforcement resources
Multiple witnesses acknowledged that existing employment laws only work if someone can enforce them. Goldman testified that the EEOC has 1,000 fewer investigators than in 1980, despite the agency now protecting 60 million more workers, receiving thousands more charges and enforcing additional laws. The Wage and Hour Division, she testified, has about 600 investigators for 165 million workers as of May 2025.
“They need to be able to hire technologists and really understand the data so they can get into the black box and do the kind of systemic work that needs to happen,” Goldman said.
Employer self-governance or blanket restrictions?
David Walton, a partner at labor and employment law firm Fisher Phillips who advises employers on AI governance, agreed that transparency and worker buy-in are critical but argued that “robust self-governance” through internal controls works better than “blanket restrictions.”
He described governance frameworks that many employers are building, such as cross-functional teams (including HR), bias audits, human-in-the-loop requirements for major decisions like hiring and promotions and training on how to report AI problems.
On the question of job displacement, Walton invoked “Jevons’ Paradox,” arguing that when technology makes a resource more efficient, it lowers costs and expands demand so much that total consumption increases. “Every automation wave that we have had, especially since 1750, has followed this pattern,” he said, predicting AI will likely increase total jobs rather than decrease them.
The committee did not announce the next steps on potential AI workplace legislation. Subcommittee Chairman Rick Allen (R-Ga.) and Ranking Member Mark DeSaulnier (D-Calif.) both emphasized the need for bipartisan collaboration. DeSaulnier noted that “history teaches us that workers don’t benefit at the same rate as investors” and urged the committee to “get ahead of that,” rather than waiting for workers to suffer.
Credit: Source link
