Which Jobs Are Most at Risk from AI?
This page answers one question: which occupations are most exposed to generative AI, and why. The ranking comes from four public datasets, task analyses by researchers and the ILO plus observed use of Microsoft Copilot and Anthropic Claude, combined into one index that we calculate and publish with its sources.
Exposure is not the same as disappearance. A high position means a large share of the role's day-to-day tasks are the kind current AI systems handle well: standardised inputs, predictable outputs, and a result someone can check quickly. It does not mean the occupation is about to end.
The occupations near the top share a shape rather than an industry. Clerical and support work, routine content production, first-line customer contact and structured analysis appear together because they rest on the same repeatable handling of documents and data, not because they belong to one sector.
Read the exposure, then read the jobs outlook. Several highly exposed occupations are still projected to grow in the US, while others are projected to shrink. Every occupation has its own page with both figures and the sources.
Top 25 Jobs Most at Risk from AI
| Rank | Job | Generative-AI exposure | US jobs 2025–35 |
|---|---|---|---|
| 1 | Call Center Agent | 99 | -5.3% |
| 2 | Customer Support | 99 | -5.3% |
| 3 | Interpreter | 98 | +2.0% |
| 4 | Translator | 98 | +2.0% |
| 5 | Web Developer | 97 | +3.8% |
| 6 | Technical Writer | 96 | +0.8% |
| 7 | Programmer | 96 | -7.3% |
| 8 | Content Writer | 95 | -0.3% |
| 9 | Copywriter | 95 | -0.3% |
| 10 | Data Analyst | 95 | +34.6% |
| 11 | Data Scientist | 95 | +34.6% |
| 12 | Software Tester | 95 | +5.7% |
| 13 | Advertising Specialist | 94 | +7.0% |
| 14 | Market Research Analyst | 94 | +7.0% |
| 15 | Marketing Specialist | 94 | +7.0% |
| 16 | SEO Specialist | 94 | +7.0% |
| 17 | Social Media Manager | 94 | +7.0% |
| 18 | IT Support Specialist | 94 | -3.5% |
| 19 | Editor | 94 | -1.1% |
| 20 | Sales Representative | 93 | -0.9% |
| 21 | Telemarketer | 93 | -21.4% |
| 22 | Data Entry Clerk | 93 | -25.5% |
| 23 | Journalist | 93 | -5.9% |
| 24 | Database Administrator | 92 | -0.1% |
| 25 | Economist | 92 | +4.7% |
Exposure runs from 0 to 100 and compares occupations with each other, using public datasets (see sources). It is not a probability that the job disappears. Methodology
Sources
Exposure index calculated by ai-job-risk.net from the datasets below (data as of October 15, 2026). US labour figures are from the BLS, unchanged.
- Eloundou, Manning, Mishkin & Rock, "GPTs are GPTs: Labor market impact potential of LLMs", Science (2024); data MIT licence
- Gmyrek, Berg, Kamiński, Konopczyński, Ładna, Nafradi, Rosłaniec, Troszyński. Generative AI and Jobs: A Refined Global Index of Occupational Exposure. ILO Working Paper 140. Geneva: ILO, 2025. © ILO, CC BY 4.0
- Tomlinson, Jaffe, Wang, Counts & Suri, "Working with AI: Measuring the Occupational Implications of Generative AI", Microsoft Research; CC BY 4.0
- Anthropic Economic Index (job_exposure); CC BY
- U.S. Bureau of Labor Statistics, Employment Projections 2025–35 and AI exposure categories
The publishers did not prepare, review or endorse this site. Full citations, licences and method
What Puts an Occupation Near the Top
Four conditions do most of the work. An occupation rarely rises on one of them alone; it rises when several hold at once.
The input arrives in a standard shape
When the same form, ticket, invoice or brief arrives thousands of times in the same format, the tool only has to be built once. Most roles high on this list spend a large part of the day moving predictable documents from one place to another.
The output can be checked in seconds
Automation spreads fastest where a wrong answer is cheap and obvious. A mis-read field or a weak first draft is caught immediately, so organisations are willing to let software try first and have a person correct it.
The volume justifies building the tool
A task has to happen often enough to be worth automating. This is why near-identical work is automated in a large contact centre and left alone in a small office: the task is the same, the economics are not.
Nobody has to be physically present
Work that is entirely digital can be moved into software without touching a building, a vehicle or a person. Once presence, physical risk or hands-on repair enters the job, the ceiling on automation drops sharply.
What This List Does Not Mean
It is not a forecast of unemployment. Occupations near the top can keep the same headcount while the content of the work changes underneath them, and some have grown even as their routine core shrank.
It is not a reading of your job. Two people with the same title can face very different exposure depending on how much of the week is routine handling and how much is judgement, relationships or physical work.
It is a snapshot of the current data editions, not a verdict. The figures change when the research and usage datasets are updated, and usage data in particular moves as adoption spreads.
Frequently asked questions
Q.Which jobs are most at risk from AI right now?
The occupations at the top are concentrated in customer contact, translation and interpreting, writing and editing, programming and testing, and clerical work. What they share is not an industry but a task profile: text and data in, text and data out, with results that can be checked quickly. The full ranking is on the rankings page, and each occupation has its own page with the sources.
Q.Does a high position mean the job will disappear?
No. This index measures how much of an occupation's work current AI systems can take on, not whether the occupation survives. In practice high exposure usually shows up first as the routine core shrinking while the remaining work shifts toward exceptions, judgement and coordination. Some occupations high on this list have kept or even grown their headcount while the content of the job changed underneath.
Q.How soon could this actually happen?
Slower than capability announcements suggest. The limiting factor is rarely whether a model can do a task; it is procurement, integration, liability and habit. Exposure is not a timetable: the BLS still projects growth to 2035 for several highly exposed occupations, and decline for others.
Q.Which jobs have already been replaced by AI?
Very few occupations have been eliminated outright. What has already happened is narrower and more specific: particular tasks inside jobs have moved to software, and some employers have stopped backfilling certain roles. We keep a separate page for the cases with documented evidence rather than mixing them into this forward-looking ranking.