Which US Jobs Are Most at Risk from AI in 2026?

The full picture: 144.0 million workers, 341 occupations

The most AI-exposed occupations in the US are clerical and administrative support roles, which score 8.5/10 - the highest of any major group. The US Bureau of Labor Statistics tracks employment across 341 detailed occupational categories covering every major sector of the economy. Using data from the BLS Occupational Outlook Handbook (OOH) May 2025 release, we scored each occupation on seven dimensions: AI exposure, robotics risk, offshoring risk, remote work potential, employment size, wage level, and projected 10-year growth.

AI exposure specifically measures how much of an occupation's core tasks can already be performed - or significantly augmented - by current AI systems. A score of 8.5/10 does not mean the job disappears tomorrow. It means a large proportion of the daily work is within reach of AI tools available right now in 2026.

The most AI-exposed occupations

Thirteen occupation groups score 8.5/10 - the highest level in our dataset. Every single one falls under the clerical and office support category. These are roles whose primary function is information handling: collecting, sorting, recording, and communicating structured data. That is precisely what large language models and AI automation tools do best.

Together these 13 occupation groups employ approximately 16.1 million Americans. That is roughly one in every nine US workers. The median wage across this group sits around $45,000 per year - solidly middle-income, not the lowest-paid work in the country. These are stable, full-time jobs that many families depend on.

Why clerical workers and not programmers?

This surprises most people. Software developers and data analysts are also exposed to AI - they score 6.5/10 in our dataset. But there is a critical difference: programmers use AI as a productivity multiplier, while clerical workers face AI as a task replacement.

A software engineer using GitHub Copilot writes more code faster. They remain employed and often more valuable. A customer service AI system handles the entire call without a human in the loop at all. The nature of the displacement is fundamentally different.

Clerical roles are also characterised by high volume, low variance tasks. They process the same types of requests repeatedly, follow established rules, and communicate in predictable patterns. AI systems trained on millions of similar interactions can match or exceed human performance on exactly these tasks.

The professionals: high exposure, high protection

The second-highest scoring major group is professionals - accountants, analysts, lawyers, and knowledge workers - at 6.5/10. This group employs around 46 million Americans and earns a median of $88,936 per year.

The high exposure reflects the reality that AI tools can already draft legal documents, produce financial analyses, write code, and summarise research. But high wages signal that employers pay a premium for judgment, client relationships, accountability, and the ability to handle novel situations. AI augments this work rather than replacing the worker entirely - at least for now.

Managers score 5.5/10 - the same as plant and machine operators. The similarity is superficial. Managers earn $118,377 median annually. Their exposure comes from AI handling administrative and analytical tasks within management. Their protection comes from accountability, organisational relationships, and decision-making in complex, human environments.

The safest jobs from AI

At the bottom of the AI exposure scale are occupations that require physical presence in unpredictable environments, direct human care, or highly specialised manual skills.

Electricians, plumbers, carpenters, and HVAC technicians score 2.5/10. They work in physical environments that change from job to job, require fine motor skills honed over years of practice, and solve problems that cannot be fully anticipated. A robot that could do this work reliably across all environments does not exist at scale in 2026.

Agricultural workers score just 2.0/10 for similar reasons - field conditions, crop variation, and physical dexterity in unstructured environments. Interestingly, their wages ($38,174 median) are lower than many high-AI-exposure clerical workers, which illustrates that AI exposure and wage level do not always align with economic risk.

What this means for you

If you work in an 8.5/10 occupation, this does not mean you lose your job tomorrow. It means the tasks you do today are increasingly within reach of AI systems, and employers will start asking whether they need the same headcount to produce the same output. The realistic timeline for significant displacement in clerical work is 3 to 7 years, not months.

The most practical response is to understand which parts of your role require judgment, relationships, and accountability - and lean into those. The workers who survive and thrive in high-AI-exposure roles are the ones who use AI tools themselves, handling more complex cases rather than the routine ones AI takes over.

If you are choosing a career or considering a change, the data suggests trades and care work are the most durable choices in terms of AI displacement risk. Electricians earn more than customer service representatives, face lower automation risk, and are in short supply. That combination rarely exists elsewhere in the labour market.

US economy and labour market context

Understanding AI's impact on US workers requires economic context. The United States is one of the highest-income economies in the world, and that shapes both the pace of AI adoption and workers' ability to adapt.

Source: World Bank Open Data (CC BY 4.0). All figures are the most recent year available per indicator.

The high GDP per capita ($90,027 in 2025, ranked among the top economies globally) means US employers have both the capital to invest in AI tools and the financial incentive to do so at scale. A US Gini of 41.8 places inequality above most of Western Europe, which matters for AI displacement: the workers most at risk from automation (clerical, lower-wage service roles) are not the same workers who have the savings, retraining time, or professional networks to absorb a rapid job change.

These composite scores are derived from World Bank economic indicators and WorldJobsData's AI disruption model. They are estimates, not official predictions, and are intended to provide directional context rather than precise forecasts.