Something shifted in March. Goldman Sachs published a report estimating that roughly 7 percent of workers will be displaced by AI. The Federal Reserve Bank of New York found that 2025 ended with the highest unemployment rate for recent college graduates in years. Amazon laid off 16,000 international roles, citing AI as the replacement. These are not anecdotes and projections any more: they are data points from institutions that typically err on the side of caution. The question of what to do about them is landing in a political vacuum.
New York Magazine's investigation into the state of U.S. policy on AI and jobs lands this week with a blunt finding: the political response has not caught up. Seventy-one percent of workers in a recent poll said they are afraid of job displacement from AI. Pro-AI political action committees are dumping millions into midterm election campaigns. And yet Congress has been, as the magazine puts it, "largely silent." The politicians New York Magazine spoke to said the issue is widely discussed in private. The absence of public proposals, they suggest, reflects something worse than ignorance: it reflects calculation.
The AI companies themselves have made more noise on the question than legislators. OpenAI, whose own optimistic projections suggest 18 percent of jobs will soon be automated, rolled out a proposal in early April for a "New Deal" for workers: a 32-hour workweek, a public wealth fund, and a tax on capital gains. Anthropic's CEO Dario Amodei described AI job disruption as "a macroeconomic problem so large" it may require a new tax code, including a levy directed "against AI companies in particular." Whether these proposals reflect genuine commitment or serve primarily to preempt regulation, nobody has been made to find out: they have so far been met with no legislative response.
The global picture adds context that the U.S. debate often omits. A study by workforce planning firm Planera, published this week, mapped automation risk across countries by sector. Malta has the highest concentration of vulnerable workers in the world: nearly half its workforce holds roles that AI can replace, concentrated in hospitality, administration, and professional services. The U.S. has the largest absolute number at risk, with an estimated 96 million workers in exposed roles. But service-dependent economies in Southern Europe face steeper proportional exposure. Greece, Spain, and similar countries where hospitality and retail dominate employment are among the hardest-hit. The international dimension matters because AI deployment is global but policy responses are national, and the most exposed economies are often those with the least capacity to respond.
One objection that has slowed the U.S. policy response is empirical: earlier projections of AI-driven job losses were often overstated, and policymakers learned to treat them with scepticism. A recent MIT study challenged the "AI-job-apocalypse narrative," but its grounds were narrow: the timeline, not the outcome. "2027 is too aggressive an estimate for AI to broadly eclipse the performance of human workers," the researchers found. "AI will achieve 80 percent success rates on most tasks by 2029." It is worth pausing over what this actually says. The researchers are not arguing the displacement won't happen; they are arguing it will happen two years later than feared. That is a thin reed on which to build a case for inaction.
Bharat Ramamurti, a senior economist in the Biden administration who has been publicly warning that AI will be "the biggest economic issue in the 2028 presidential election," told New York Magazine he expected politicians to move on the issue sooner. The gap between the scale of what analysts are projecting and the volume of legislative activity is unusual even by Washington's standards. What it suggests, among the politicians the magazine interviewed, is that the disruption is real enough that they see it coming, but not yet politically acute enough to reward bold proposals. The calculus may change quickly: the data is getting harder to explain away, the midterms are approaching, and the AI companies are already spending heavily to shape whatever comes next.