A US Senator Wants Laws to Rein In AI Before the Damage Gets Worse
Senator Ed Markey has introduced a package of bills targeting AI's biggest real-world harms: water-hungry data centres, biased hiring algorithms, and workers being overruled by software they never saw.
Key points
- Senator Ed Markey, Democrat of Massachusetts, has proposed a package of new laws targeting specific harms caused by artificial intelligence systems.
- The bills take aim at large data centres, which consume enormous amounts of water and electricity to keep AI running.
- One proposal targets automated hiring tools, which screen or reject job applicants without a human making the call.
- Markey's package also addresses algorithmic bias, where software trained on flawed historical data can discriminate against workers or customers.
- The legislation arrives as a handful of technology companies report record profits from AI products while wider economic inequality grows.
Senator Ed Markey is not worried about science-fiction robots. He is worried about what is happening right now, in offices, warehouses, and server rooms across the United States.
Markey, a Democrat from Massachusetts, has introduced what he calls an "AI accountability agenda": a set of bills designed to put legal guardrails around artificial intelligence before its harms become too entrenched to fix. The Guardian Technology first reported the details of the package.
What would these bills actually do for ordinary people?
They would give workers and consumers legal protections that do not currently exist. Take automated hiring systems, software that reads your CV and scores or rejects you before a human ever sees your name. Markey's proposal would force companies to disclose when they use these tools and give applicants a way to challenge a decision made by an algorithm, a set of rules a computer follows to reach a conclusion.
Then there is the data centre problem. A data centre is a large building packed with servers, the physical computers that run AI software. These facilities use staggering amounts of electricity and water, the water mostly for cooling. Markey's bills would require greater transparency about that environmental footprint, putting the costs on the balance sheet where the public can see them.
The failure mode here is familiar: a technology scales faster than anyone's ability to measure its side effects, and by the time the measurement happens the side effects are somebody else's problem.
Workplace surveillance is on the list too. More employers now use AI to monitor how fast workers move, how long they pause, even the tone of their voice on calls. Markey's package would restrict how that data gets collected and used.
The economic inequality angle is the sharpest edge. A small number of companies are capturing most of the financial gains from AI. In practice, that means the workers whose jobs AI changes or eliminates see little of the upside. The bills do not ban AI profit, but they push back against a model where the windfalls flow one direction only.
One thing the post-mortem on this legislative moment will say is that the window for sensible rules was narrow and most of it passed quietly.
If you work in a job where software rates your performance or screens your applications, check whether your employer publishes a policy on how those tools work. If they do not, ask.
