Call centres, clerks, receptionists, and accountants top the most exposed list, and women and graduates carry that exposure because those occupations have sorted them in for decades.
Australia's first national audit of which jobs AI can take over names a specific list: telemarketers, call-centre workers, clerks, retail managers, software programmers, accountants, receptionists, and advertising and marketing professionals sit at the top of the most-exposed rankings. Tradespeople, aged-care workers, carers, truck and forklift drivers, cleaners, and gardeners sit at the bottom. The Department of Employment and Workplace Relations released the report and frames it as a task-exposure map, not a layoff forecast. The list alone is the news. The demographic pattern that falls out of the list is what the next week of Australian AI policy turns on.
The most-exposed jobs are dominated by routine cognitive work: scripted replies, document summaries, scheduling, reconciling figures, processing sales. These are the same entry-level white-collar roles Anthropic chief executive Dario Amodei told The Guardian could be wiped out within one to five years, with unemployment rising to 10–20% along the way. Australia's classification lines up with that framing. It also produces a recognisably Australian demographic story. Call centres, retail management, reception work, clerk-grade offices, and marketing support have been heavily female workforces for decades, and they sit above the national average for tertiary attainment. The least-exposed end skews toward vocational qualifications: apprenticeships, care certificates, heavy-vehicle licences, cleaning and grounds tickets.
Women and university graduates are not more AI-vulnerable because of who they are. They are more AI-vulnerable because that is where the labour market has sorted them. The report's exposure score, drawn from Jobs and Skills Australia (JSA) occupational data published alongside the DEWR audit, measures the share of an occupation's current tasks that current generative AI tools can plausibly substitute or augment. A high score means the work is exposed. It does not mean the role is being cut. No Australian dataset yet shows sustained AI-driven displacement at occupation level, and employment minister Amanda Rishworth told the Guardian that labour-market conditions remain strong by historical standards and youth outcomes have mostly held up.
The report's headline risk language is "most exposed", not "losing jobs". The bolder displacement claim is coming from outside the audit, mostly from the CEO who has the most to gain from Australia buying AI infrastructure. Anthropic is pursuing multibillion-dollar investment in the country, and its economic index is the analytical scaffolding for the call-centre classification inside the report. A vendor warning backed by the vendor's own modelling is not a consensus forecast. Framing the JSA score as "imminent job loss" treats exposure as displacement and the vendor projection as baseline.
The Albanese government is expected the following week to release updated AI regulation and management plans covering industry, the economy, and safety guardrails. The exposure map lands in the same news cycle. Read at sector level, it suggests that any retraining and transition scheme aimed at "people most at risk" needs to start from sectors, not from individual credentials. Staffing patterns inside contact centres, retail chains, clerical back offices, and marketing functions are what produce the demographic pattern, and procurement rules, contracting norms, and award coverage are what keep those patterns in place. The report calls for ongoing tracking, not immediate intervention, which leaves the policy question as the more important of the two for now.
Australia has not previously tracked AI occupational exposure at national scale, so this is also the dataset that future editions of the audit will measure against. Today's exposure map is a snapshot of where current AI tools can do current tasks. The next edition will be the first one able to test whether exposure scores actually track to employment movement.