Two visions of AI's future collided in Washington this week, and both are in some sense responses to the same problem. Senator Bernie Sanders and Representative Alexandria Ocasio-Cortez introduced legislation to halt new data center construction until Congress passes comprehensive AI regulation. One day earlier, Bloomberg's opinion desk published a pointed assessment of the Trump administration's own AI framework: that it is less a plan than a list of things the administration hopes someone else will do.
The Sanders-AOC bill is more ambitious than the name suggests. It does not simply pump the brakes on construction. It calls for government review and certification of AI models before release, mandatory protections against AI-driven job displacement, limits on the environmental footprint of AI infrastructure, and union-labour requirements on data center construction projects. The bill frames AI as a labour and environment problem as much as a technology one. The data center construction ban is the headline, but the job-displacement provisions are arguably the more consequential ask.
The bill has essentially no chance of passing in the current Congress, and everyone involved almost certainly knows it. Its real function is to establish a counter-narrative: that the federal government's current posture of deregulation and state-law preemption is a choice, not a default, and that alternatives exist. Progressive politicians have been watching the Trump administration try to consolidate AI policy authority at the federal level while simultaneously stripping back regulation, and the data center bill is partly a response to that manoeuvre.
The Trump framework, meanwhile, drew a sceptical reading from Bloomberg's opinion desk. The administration's six-pronged outline covers everything from child-safety rules to data center permitting to energy use standards. The White House says it wants the framework codified into law before the end of the year. Bloomberg's view is that the plan does very little that the administration actually controls: it asks Congress to pass legislation, asks states to stand down on their own rules, and asks industry to behave responsibly. What it does not do is impose obligations on anyone.
This puts the two sides in an odd symmetry. The progressive bill demands much more than it can deliver. The White House framework delivers less than it promises. Both are staking out political positions more than drafting governance. The actual regulatory vacuum in the middle, where questions about model liability, AI-driven hiring decisions, and data centre permitting actually need answers, remains largely unaddressed.
The state-level tension cuts across both. California, New York, and others have been moving ahead with their own AI legislation, partly because federal action has been so slow and so inconsistent. The Trump administration has threatened to withhold federal infrastructure funding from states that pursue AI rulemaking, a move that has not deterred California legislators. The Sanders-AOC bill, interestingly, would reassert federal jurisdiction, but from the opposite direction: rather than preempting state rules with a light-touch federal framework, it would preempt them with a more restrictive one.
What makes this week's confluence interesting is that both the White House framework and the progressive bill take the data center buildout as their organising fact. The question is not whether to build at scale: hundreds of billions of dollars in commitments are already in motion. The question is who sets the conditions. As Bloomberg noted, the administration's plan mostly assumes the industry will show up to negotiate in good faith. Whether that assumption holds, with so much capital already committed and so little federal enforcement mechanism in place, is the open question neither document really answers.