The first official report from the Australian Department of Employment and Workplace Relations on the impact of artificial intelligence on the labor market did not reveal mass layoffs, but it did record a distinct trend: employment growth in professions most vulnerable to automation is significantly lagging behind other segments. This is an early but important signal that generative AI is already changing the structure of labor demand.
According to an analysis of data from November 2022 to February 2026, in the fifth of professions with the highest exposure to generative AI, employment grew by only 5.6%. For comparison, in the group least affected by the technology, this figure was 9.5%. The nearly twofold difference is not a coincidence but a systematic slowdown, which the report's authors call a "moderate but real" signal.
The department's model shows: for a profession with an AI exposure level one standard deviation above the average, employment by February 2026 turned out to be approximately 2% lower than would have been expected if the pre-pandemic trend had continued. However, the result is sensitive to methodology. When using alternative exposure metrics or excluding the COVID-19 period, the statistical significance of the negative relationship disappears. This suggests that we are observing not an explosive collapse, but a gradual shift.
Who is at risk: offices under threat, hands-on jobs safe
The most vulnerable turned out to be routine cognitive professions: office workers, data entry operators, administrators, accountants, advertising and marketing specialists, as well as programmers. The least exposed to AI are professions involving manual labor and care: caregivers, electricians, drivers, cleaners, carpenters, plumbers, and gardeners. This confirms a shift in the classic view of automation: previously, physical labor was under threat; now, it is intellectual labor.
Particular attention is drawn to the gender and educational gap. In the high-risk group, men make up only 43.7% of workers, whereas in the least vulnerable group, they account for 69.5%. The share of workers with a bachelor's degree in the first group reaches 43.7%, compared to 14.9% in the second. Women and university graduates are disproportionately represented in professions where tasks are easier to automate.
At the same time, the picture is not simply one of displacing "white-collar" workers. Software and application developers in Australia numbered 199,000 by February 2026 — 25% higher than the November 2022 level. The market is not destroying these professions but is redistributing demand within them.
No mass upheaval — but the trend is clear
Australia's unemployment rate in February 2026 was 4.2% — lower than any value in the decade before COVID-19. The employment-to-population ratio reached 64%, above the average pre-pandemic level. More noticeable signals appeared precisely in the group of highly exposed professions: there, employment, hours worked, and the number of vacancies grew more slowly, while unemployment increased more sharply. However, the report's authors emphasize that many routine cognitive professions were declining as a share of employment even before the advent of ChatGPT. The current weakness may reflect not only AI but also post-pandemic restructuring and macroeconomic factors.
The report presents a wide range of forecasts — from moderate to catastrophic. Anthropic CEO Dario Amodei, for example, predicts that AI could eliminate up to half of entry-level "white-collar" jobs and push unemployment to 10–20% within one to five years. These scenarios have not yet materialized, but the direction of movement has already been set.
The Australian government is preparing a new approach to AI regulation, including issues of safety, economics, and copyright. Minister for Employment Amanda Rishworth stated that the goal is to use AI to create "good jobs, not to threaten them." The department plans to continue quarterly monitoring, although the next update may only come at the end of 2026 due to a transition to a new occupation classification.
My analysis: Australia's data is not panic, but an important empirical signal. The labor market is indeed beginning to adapt to generative AI, not through layoffs, but through a slowdown in hiring. The key risk lies not in short-term unemployment, but in long-term structural instability for cognitive professions. Investors and analysts should watch how companies reallocate their personnel budgets: this will become an indicator of which sectors will gain and which will lose in the next 3–5 years.