US vs World: How Does America's AI Job Risk Compare in 2026?

When AI dominates headlines, a natural question emerges: is the risk the same everywhere, or are some countries more exposed than others? We scored the workforces of five OECD economies - the United States, United Kingdom, Germany, Australia and Canada - using official government labour statistics and the same scoring methodology across all five. The result is 268 million workers, one comparable dataset.

The headline answer is that the differences are smaller than most people expect, but they are real - and the reasons why matter as much as the numbers themselves.

The Scoreboard: Five Countries, Four Dimensions

Before diving into the analysis, here is the full comparison across the four dimensions we track: AI exposure, risk velocity, recovery resilience and sovereign buffer.

AI exposure is a weighted average across all occupation groups, where each group is weighted by its share of total employment. Scores run 0-10. Velocity, resilience and buffer are composite scores derived from labour market, economic and policy data. Higher resilience and buffer = better prepared.

Why Germany Scores Highest

Germany's top ranking on AI exposure (5.30/10) is not simply because its workers face more AI - it is because the German data is more granular. While the US dataset uses 9 broad ISCO occupation groups, Germany provides data at 41 sub-groups. This reveals pockets of extreme exposure that get averaged away in broader datasets.

The standout numbers:

• IT professionals: 8.5/10 - highest of any professional group across all five countries. Germany's export-driven economy has a large engineering and software workforce, a group now deeply disrupted by AI coding tools.
• Business and administration professionals: 8.0/10 - finance, accounting, HR, legal support. A massive white-collar middle layer.
• General and keyboard clerks: 9.0/10 - the highest single score in any of our five country datasets. Germany's Mittelstand model relies heavily on administrative coordination that AI is actively targeting.

The US and UK: A Statistical Tie With Very Different Stakes

The United States (5.07/10) and United Kingdom (5.08/10) are separated by a rounding error on AI exposure. But their situations are meaningfully different on the two dimensions that determine how badly disruption hurts.

Both countries score 10.0/10 on risk velocity - meaning AI adoption is happening now, not in some distant future. But the UK's sovereign buffer score of 1.3/10 is the weakest of any country in our dataset. This reflects a combination of stretched public finances, limited retraining programme capacity, and a workforce that is more concentrated in London-based financial and professional services - precisely the sector most exposed to AI.

The US at 2.1/10 is not far ahead, but its sheer economic scale, larger private sector retraining market and more distributed geography provide some additional cushion the UK does not have.

The Occupation-by-Occupation Breakdown

Across all five countries, the ranking of occupation groups by AI exposure is remarkably consistent. Here is the full comparison at the broad ISCO-1 level:

Where the US Has an Edge - and Where It Doesn't

The US often leads global rankings on technology adoption, but that cuts both ways on AI job risk. The US has more technology workers exposed to AI coding tools, more financial professionals exposed to AI analysis, and more administrative workers exposed to AI document automation. This is why velocity sits at 10.0/10 - AI is already being deployed at scale in the US, faster than in any of the other four countries.

The US advantage: wages and retraining runway

Where the US genuinely differentiates is on the financial position of displaced workers. US clerical workers earn $45,433/year on average - well ahead of UK clerical workers at $33,216. That income gap translates directly into more months of savings to fund retraining or job searching during a transition. Higher wages in at-risk roles mean individual workers have more capacity to absorb disruption, even if the state cannot fund it.

One result here stands out: Germany pays its clerical workers slightly more than the US ($46,899 vs $45,433) despite a lower national average wage. This reflects Germany's strong collective bargaining agreements and sectoral wage floors - a structural protection that partially offsets the exposure risk.

The US disadvantage: the weakest safety net

The sovereign buffer score is where the US falls behind. At 2.1/10, the US has a weaker fiscal and policy capacity to manage mass displacement than Australia (6.1/10), Canada (5.7/10) or Germany (which has strong labour market protections even without a scored buffer). The American model places the burden of retraining almost entirely on individual workers and employers, with limited national-level programmes equivalent to Germany's Kurzarbeit short-time work scheme or Australia's Jobs and Skills Australia framework.

Australia: The Quiet Outlier

Australia's position is counterintuitive. At 4.95/10 AI exposure - the lowest of the five countries - it appears safest. But this is partly a data artefact: Australia's workforce has a proportionally larger share of healthcare, construction and resources workers relative to population, all of which score low on AI exposure.

What makes Australia genuinely different is its preparedness. At 7.8/10 recovery resilience (tied with Germany) and 6.1/10 sovereign buffer (highest in our dataset), Australia has more capacity to fund and execute worker transitions than any other country we have analysed. Its smaller total workforce (17.0M vs 155.5M in the US) also means displacement can be addressed at a more manageable scale.

Canada: The Hidden Middle Child

Canada sits between the US and Germany on AI exposure (5.29/10) and has a risk velocity of 9.2/10 - fast, but slightly below the US and UK. What makes Canada notable is its offshoring risk: Canadian professionals score 6.0/10 on offshoring exposure, meaning they face not just AI displacement but a parallel threat from global remote work shifting jobs to lower-cost locations.

Canada also has the lowest average clerical wage of the five countries at $36,940 - lower even than the UK ($33,216 in USD terms is close, and both are below the Canadian figure when adjusted). Workers in the highest-risk group have less financial runway than their American or Australian counterparts.

The One Finding That Unites All Five Countries

Amid the differences in scores and structures, one finding holds true across every country we have ever analysed: tradespeople are the most protected occupation group from AI.

Electricians, plumbers, carpenters, pipe fitters and HVAC technicians score between 2.0 and 2.5/10 on AI exposure in every country in this comparison. Physical dexterity in variable environments, real-time adaptation to unstructured problems, and hands-on safety judgements remain deeply difficult for AI systems. The gap between a clerical worker (8.5/10) and a trades worker (2.5/10) is not narrowing - it is widening as AI accelerates into knowledge work while leaving physical work largely untouched.

This matters for individual career decisions, education policy, and government retraining programmes in every country on this list.

What Should Workers in Each Country Do?

The country-level picture is useful for policy. But if you are an individual worker, the relevant question is where your occupation sits, not your country. The table above shows that a clerical worker in Australia faces almost identical AI exposure as a clerical worker in Germany - the country of residence changes the safety net, not the risk.

For each of the five countries, we have published a full breakdown by occupation with wages, projected growth, and AI rationale for each score:

Methodology Note

All AI exposure, robotics risk, offshoring risk and WFH potential scores are derived from the WorldJobsData scoring model, which draws on the Frey-Osborne (2013) Oxford automation probability research, OECD task-content analysis, and ILO ISCO-08 occupation classifications. Employment data is sourced from national statistics offices: BLS (US), ONS (UK), Destatis (Germany), ABS (Australia), Statistics Canada. OECD average annual wage data is from the OECD Employment Outlook 2024.

Weighted averages weight each occupation group's AI exposure score by its share of total national employment. This means large groups (e.g. US Professionals at 30% of employment) have more influence on the national average than small groups (e.g. Agriculture at 0.6%).