The White House Says AI Is Costing Nobody Their Job

White House National Economic Council Director Kevin Hassett told CNBC on Monday that there is no sign in the data that AI is costing anyone their job right now. The comment landed on the same day that the cumulative AI-linked tech layoff count for 2026 stood above 92,000. The distance between those two positions is not merely a matter of interpretation. It is a reflection of two fundamentally different ways of reading the same labor market, and the leadership implications of choosing one over the other are significant.

‍

What Hassett Said and What the Data Shows

‍

Hassett's position has internal logic. Aggregate unemployment remains at 4.3%. The April jobs report beat expectations, adding 115,000 nonfarm payroll jobs. No single data series yet shows a spike in unemployment attributable cleanly and directly to AI-driven displacement. In that narrow framing, the statement is technically defensible.

The problem is that the framing is too narrow. Research on AI labor market effects consistently identifies the first phase of disruption, spanning roughly 2023 to 2025, as one characterised by task automation, hiring freezes, and role compression rather than mass termination. Companies integrate AI into existing workflows, reducing the need for new hires rather than firing existing workers at scale. AI suppresses hiring more than it destroys jobs. That effect does not show up cleanly in layoff statistics or unemployment claims, because the jobs it eliminates are the ones that were never posted.

An unemployment rate that holds steady while hiring volumes collapse is consistent with AI displacement. It is also what the data shows. The Bureau of Labor Statistics reports that the 12-month average payroll gain, even accounting for an unusually strong March, sits around 22,000 jobs per month. Excluding healthcare, driven by demographic demand the technology has not yet meaningfully penetrated, the economy has seen a net loss of positions over the past year.

Anthropic's research team, developing a measure of AI displacement risk based on actual usage data rather than theoretical exposure, found evidence that high AI-usage occupations are beginning to see modestly slower hiring. The effects remain modest but are accelerating. That is a different signal from no sign in the data.

‍

What the Companies Are Saying

‍

The gap between Hassett's characterisation and the public statements from corporate leadership teams doing the actual cutting is not subtle.

‍

Block cut nearly 4,000 employees in February, reducing its headcount by half. CFO Amrita Ahuja stated plainly that the company was choosing to shift how it operates by using AI to automate more work. Coinbase cut 14% of its workforce on May 5. CEO Brian Armstrong described watching engineers use AI to ship in days what used to take a team weeks, and non-technical teams shipping production code, as the direct rationale for needing fewer people. Cloudflare cut 1,100 jobs, 20% of its workforce, citing agentic AI as having fundamentally changed how the firm operates, with internal AI usage up more than 600% in three months. Atlassian eliminated 1,600 roles to self-fund further investment in AI. Oracle has reduced its headcount substantially while committing $156 billion to AI infrastructure. Meta is cutting 8,000 employees by May 20 while simultaneously grading all remaining employees on their AI usage.

Each of those announcements explicitly connects headcount reduction to AI-enabled productivity gains. These are not companies citing a difficult market or declining revenue. Several of them are reporting record earnings while making the cuts. The mechanism is the one Hassett's framing does not capture: companies are discovering they can generate the same or greater output with fewer people because AI is handling a growing share of the work. The people who were doing that work are losing their positions.

Out of confirmed tech layoffs through early 2026, approximately one in five were explicitly attributed to AI and automation by the companies announcing them, a dramatic increase from 2025 when AI was cited as a factor in fewer than 8% of layoff announcements. The trend is accelerating, not stabilising.

‍

The Hiring Suppression Effect

‍

The most important mechanism Hassett's statement overlooks is one that never generates a layoff announcement, a severance payment, or an unemployment claim. It is the job that is simply never opened.

‍

Goldman Sachs research identifies hiring suppression as the primary channel through which AI is currently affecting employment. A company that previously would have added 20 analysts to handle growing data volumes instead deploys an AI system and adds two. The 18 positions that were never created do not show up in layoff data. They do not generate unemployment claims. They do not move the headline unemployment rate. They are invisible to the metrics Hassett is reading.

Job postings for roles involving structured and repetitive tasks decreased by 13% since the public launch of ChatGPT in late 2022, while demand for roles requiring analytical, technical, or creative work grew by 20%, according to Harvard Business School research. The overall job count looks stable. The composition of available jobs is shifting substantially, and the workers whose roles are being automated away are frequently not the ones qualified for the roles being created.

The Federal Reserve Bank of Dallas, reviewing wage and employment data through early 2026, found that in AI-exposed occupations, wages are not uniformly declining, suggesting that for many workers AI is currently augmenting rather than replacing output. The research consensus is that the most significant labor market effects of AI will materialise between 2027 and 2030, as current deployments mature, autonomous systems reach commercial scale, and productivity effects accumulate across knowledge work.

‍

That finding supports a version of Hassett's argument: the severe displacement has not yet fully arrived. What it does not support is the characterisation that there is no sign in the data. The signs are present. The full effect is still building.

‍

The Leadership Responsibility in the Gap

‍

For CEOs, CHROs, and board members, the question Hassett's statement raises is not primarily political. It is a governance question about how corporate leadership accounts for the human consequences of the workforce decisions they are making.

The companies cutting thousands of jobs while citing AI productivity are making economically rational choices within their own P&L frameworks. The case for doing so is straightforward: AI tools are available, they work, they reduce the headcount required to produce a given output, and shareholders expect management to optimise accordingly. Nothing in that logic is dishonest.

What several of those companies have been less explicit about is the transition cost, borne primarily by the workers affected rather than the companies making the decision. Among roughly 37 million highly AI-exposed U.S. workers, approximately 70% have sufficient adaptive capacity and transferable skills to navigate a role shift if their current positions change. The 30% who lack that capacity represent approximately 10.6 million workers, concentrated in lower-wage roles, with fewer retraining resources, in geographic areas with fewer alternative opportunities. That group is not served by reassurances that the data shows no sign of AI job losses.

The World Economic Forum's Future of Jobs Report projects that by 2030, 85 to 92 million jobs will be displaced globally while 97 to 170 million new roles emerge. The net figure is positive. The transition period between displacement and re-employment is where the human cost concentrates, and that transition period is now.

Hassett is correct that a large-scale collapse in employment has not yet materialised. The data on what is actually happening in corporate hiring rooms, in the job postings that are being pulled rather than filled, and in the layoff announcements that explicitly cite AI productivity as their rationale, tells a more specific story than aggregate unemployment can capture. Leadership teams that are making workforce decisions based on the clearer picture are doing so with a fuller understanding of what they are choosing and who bears the cost of it.

‍

The White House has a task force studying the future of AI and the workforce. The companies already restructuring around it are not waiting for the study to finish.