G7 AI Job Risk Ranked: The US Isn't Even #1

The full G7 ranking

AI exposure scores are employment-weighted averages: every occupation group's AI exposure score is multiplied by its share of the national workforce, then summed. A country scores higher when a larger share of its workers sit in clerical, professional, and technical roles - not because AI itself works differently there.

Source: ILO ILOSTAT employment-weighted occupation data, each country's own site/data/countries/[code].json. Recomputed and independently verified against each country's stats.weighted_avg_ai_exposure field, 2026-09-24.

Why Germany beats the US - one occupation group explains most of it

Germany's clerical support workers score 8.5/10, the same as the US. Its professionals score 7.0/10 versus the US's 6.5/10 - a real but modest difference. The gap that actually separates the two countries is technicians and associate professionals: in Germany, this group is 20.1% of the workforce and scores 6.1/10. In the US, the same group is only 5.8% of the workforce, scoring 5.5/10.

That one category - engineering technicians, quality control associates, IT support and operations roles - is more than three times larger as a share of the German workforce than the American one, and it scores higher per worker. Germany's dual vocational training system (Ausbildung) channels a much larger proportion of young workers directly into technician-tier roles than the American labor market does, where the same functions are more often split between a smaller technician tier and a much larger professional tier that absorbs people through four-year degrees instead.

The result: Germany doesn't have more advanced AI, and its clerical workers aren't more exposed than America's. It simply has more of its workforce sitting in the specific occupation tier where AI exposure is moderately-high and the group is large - a direct, measurable structural difference, not a technology gap.

Canada's case is different - and closer to Germany's than the US's

Canada's path to 5.29/10 doesn't run through technicians the way Germany's does - Canada's technician share (19.7%) is close to Germany's, but its real edge over the US comes from clerical support workers: 12.6% of the Canadian workforce versus 11.3% in the US, both scoring 8.5/10. Canada's professional share (23.9%) is actually slightly lower than the US's own (29.6%), scoring the same 6.5/10 - so the professional tier alone would predict the US scoring higher, not lower.

What flips the result is that Canada's workforce is proportionally thinner in the very-low-exposure tiers - Canada has less of its labor force in occupations that pull an average down, so its clerical and technical concentration carries more weight in the national score, even with a smaller professional class than the US.

Speed and exposure are different questions - the US leads one, not the other

Occupation-level exposure answers "how much of this country's work looks like what AI is good at." Risk velocity answers a different question: how fast will that exposure actually convert into deployed AI tools, given digital infrastructure, enterprise AI adoption rates, and broadband penetration. On velocity, the US and UK both score the G7 maximum of 10.0/10 - the fastest deployment timelines in the group, even though neither country has the highest occupation-level exposure.

Germany and Canada, despite scoring higher on exposure, sit slightly below on velocity (9.6 and 9.2 respectively) - still near-maximum, but not tied for first. The practical read: Germany and Canada have more of their workforce structurally positioned in AI-exposed roles, but the US and UK are moving fastest to actually deploy the tools that act on that exposure. Both facts are true and they answer different questions - "how exposed" and "how soon."

Italy: lowest exposure, but the one G7 country with a real resilience gap

Italy scores lowest in the G7 on AI exposure (4.82/10) - modest comfort, but worth pairing with its recovery resilience score of 5.8/10, the only G7 country rated medium resilience rather than high. Every other G7 country in this ranking - including the US, which has among the weakest formal labor protections in the group - scores 7.0 or above on recovery resilience, reflecting relatively strong institutional capacity to retrain and redeploy displaced workers even where exposure is high.

Italy's combination - lower exposure but weaker institutional capacity to absorb the disruption that does happen - is a genuinely different risk profile than "low exposure, therefore low risk." A smaller share of Italian workers face AI pressure, but the ones who do have less structural support than their counterparts anywhere else in the G7.

What this means if you work in a G7 country

If you're in clerical or administrative work in Germany, Canada, or the UK, the occupation-level data says you're at least as exposed as an equivalent worker in the United States - the "America is ground zero for AI job loss" framing that dominates most coverage doesn't hold at the occupation level. Clerical support scores 8.4-8.6/10 in every single G7 country; there is no G7 economy where this work is meaningfully protected.

If you're a technician or associate professional in Germany specifically, your occupation tier is both unusually large and moderately exposed (6.1/10, 20.1% of the workforce) - a combination that makes this group worth watching closely in Germany in a way it doesn't apply the same way to peer countries, where the technician tier is smaller.

If you're in the US or UK, the practical planning horizon is shorter than the exposure score alone implies, because both countries lead the G7 on deployment speed. A moderate exposure score with maximum velocity still means near-term disruption for the roles that are exposed - the score tells you how much of your occupation is exposed, the velocity tells you how soon that exposure becomes an actual tool your employer is using.