Senator Ed Markey, a Democrat from Massachusetts, has unveiled what he calls an "AI accountability agenda," a package of proposed bills targeting the real-world effects of AI on ordinary people. The package covers data centers, workplace AI, discriminatory algorithms, child safety, and healthcare. None of this is law yet. Congress has passed exactly one AI-specific federal law so far: the TAKE IT DOWN Act, signed in May 2025, which targeted non-consensual intimate images. Everything else is still being negotiated, blocked, or ignored.
The political gap between Markey's proposals and what is actually moving through Congress is wide. The Trump administration has taken the opposite direction: its December 2025 executive order pushed to limit state-level AI regulation, created a task force to legally challenge state AI laws it deems too strict, and positioned AI primarily as a national competitiveness issue. The White House released a non-binding national AI framework in March 2026 that explicitly recommended Congress not create any new federal AI regulator.
That said, one of Markey's recent victories is notable. In July 2025, the Senate voted 99 to 1 to strip a proposed ten-year ban on state AI regulation from a Republican budget bill. That bipartisan vote suggests there is appetite, even among Republicans, for states to retain the power to protect consumers, even if there is no consensus on federal rules.
The electricity cost issue is where Markey's agenda has the most immediate relevance for business operators. Data centers that power AI tools now consume enormous amounts of electricity, and that cost is flowing to everyone else. US residential electricity prices rose 11.5% in 2025, outpacing inflation. In the northeast, the main electricity grid operator reported that data center demand drove capacity costs up by roughly $9.3 billion for the 2025-26 delivery period. Utilities across the country requested more than $29 billion in rate increases in just the first half of 2025. Those costs land on whoever pays an electricity bill, whether a household or a small business.
Markey's proposed data center certification bill would require companies to get federal approval before building new facilities, with reviews covering air and water quality, noise, energy costs, and local economic impact. This would create a new regulatory step that does not currently exist. Construction timelines for data centers would likely lengthen if this passes, which matters for any business dependent on cloud services or AI infrastructure.
The workplace proposals are more immediately relevant for HR and operations managers. In 2024 alone, AI-powered hiring tools processed over 30 million job applications in the US while also generating hundreds of discrimination complaints. Markey's bills would ban employers from primarily relying on automated systems for hiring, firing, and promotion decisions, and would protect workers who disagree with AI recommendations. A separate bill would restrict broad digital surveillance of employees.
Some of this is already moving at the state level regardless of what Congress does. California, Illinois, Colorado, and New York City already have active requirements around AI tools in hiring. Employers in those states must conduct bias audits, notify candidates, and maintain documentation. New York City fines for violations can escalate into the millions. Any business with US operations that uses an AI tool to screen applicants or track staff performance needs to understand where it stands under current state law, not just federal proposals.
The pattern here is clear: federal law is slow and contested, but state law is moving fast and in multiple directions at once. Markey's bills represent one pole of that debate. The White House represents another. Companies operating across multiple US states face a patchwork of requirements that are only going to get more complex over the next two to three years. The businesses that will handle this best are the ones that start documenting their AI use now, before the rules arrive.