Around 1.4 million women in Australia work in clerical and administrative roles, and the vacancy data for June 2026 confirms what the exposure modelling had been flagging for months: displacement is already underway. Jobs and Skills Australia (JSA), the federal government agency that calculates an automation exposure score for every occupation in the workforce, found that 15 of the 20 roles most at risk of AI displacement are female-dominated. The story is no longer a medium-term forecast. It is showing up in this quarter's hiring numbers.

Clerical and administrative work is more than 70% female in Australia and accounts for close to one in five jobs held by women across the entire workforce. The roles sitting at the intersection of high AI exposure and high female concentration include secretaries, receptionists, bookkeepers, accounting clerks, HR clerks, and payroll clerks. That is not a narrow slice of the economy. It is a structural seam, and the vacancy data shows it is thinning.

What the June 2026 vacancy data shows

Job openings for personal assistants and bookkeepers in June 2026 were around 22% lower than at the same point a year earlier. Openings for general clerks were down around 9%. Across the broader Australian labour market, total vacancies fell by only around 1.7% over the same period. This is not a general hiring slowdown driven by interest rates or consumer confidence. It is a targeted contraction concentrated in the roles generative AI absorbs most cheaply.

The contrast with less-exposed occupations is stark. Over the same window, vacancies for concreters rose 5.5% and electricians climbed 14.3%. Of the 20 occupations least exposed to automation, 17 are male-dominated. Of the 20 most exposed, only one, financial dealers, is even moderately male-dominated. The rest of the top 20 are staffed predominantly by women, and they are the roles losing job openings fastest right now.

The Australian Services Union (ASU), which covers clerical, administrative, and community sector workers, has long argued that the undervaluation of administrative work leaves women in these roles with fewer buffers when conditions shift. The June 2026 vacancy numbers give that argument a harder edge.

For employees: what this means if you work in admin

If your job title is receptionist, secretary, personal assistant, bookkeeper, or HR or payroll clerk, the exposure score is not saying your role disappears tomorrow. It is saying the tasks that fill most of your day, drafting standard correspondence, coordinating calendars, reconciling ledgers, running first-pass CV screens, processing payroll cycles, are precisely the tasks generative AI handles at the lowest marginal cost. That gap will close faster than most workforce plans currently assume.

A few concrete moves worth prioritising over the next six months, in order of likely impact:

• Audit your own tasks against AI tools. Identify the five things that consume most of your working week. For each, test whether a mainstream AI tool can replicate the output. The exercise surfaces two useful facts: which parts of your role are already automatable, and which parts still depend on your judgement. Both are worth knowing before your next performance review.
• Shift your centre of gravity inside the current role. A payroll clerk who becomes the person reviewing AI-generated payroll runs for anomalies and escalating exceptions is harder to replace than one who runs the cycle manually. Same job title, materially different value to the employer.
• Look sideways before looking up. Adjacent roles involving client relationships, exception handling, compliance judgement, or people management carry lower exposure scores than pure processing work. A bookkeeper moving into management accounting, or an office manager stepping into an operations coordinator role, is following the exposure gradient in a productive direction.
• Get the salary conversation on the record now. If you are being asked to supervise AI-generated outputs that previously required a colleague, that is a scope change worth documenting. Benchmark your pay against current market rates before your next review, not after. The wherewework salary data tool can give you a verified reference point drawn from people in comparable roles.

One caveat worth naming plainly: the roles contracting fastest are also the roles most women use to enter the workforce, return after a career break, or structure work around caring responsibilities. The Fair Work Act 2009 (FW Act) provides minimum entitlements including flexible work request rights under the National Employment Standards (NES, Part 2-2 of the FW Act), but the NES does not protect a role from disappearing through attrition. The realistic move for most people in this situation is lateral, into roles that still draw on clerical skills but layer in judgement, client relationships, or specialist domain knowledge. Retraining into a trades occupation is a policy-level answer, not a practical one for most people managing real life.

For HR and recruitment leaders: the pipeline is already shifting

The 22% year-on-year drop in personal assistant and bookkeeper vacancies is not a projection about 2030. It is your June 2026 hiring pipeline. If your workforce plan still treats clerical and administrative hiring as a stable baseline category, the plan is operating on assumptions the market has already moved past.

The Australian HR Institute (AHRI) has consistently flagged workforce planning capability as one of the largest gaps in Australian HR practice. The JSA vacancy data gives that gap a specific, measurable shape. Three items worth putting on the agenda for the next planning cycle:

1. Run a gender-aware exposure audit of your own workforce

The Australian federal government already requires all new policy proposals to undergo a gender analysis. The researchers behind the JSA automation exposure work argue this requirement should extend to the newly established Office of AI. The same logic applies at the enterprise level. Map your clerical and administrative headcount by gender. If 70% or more of the roles you are quietly not replacing are held by women, that pattern will surface in your Workplace Gender Equality Agency (WGEA) reporting whether you have planned for it or not. Under the Workplace Gender Equality Act 2012, employers with 100 or more staff report annually to WGEA on gender composition and remuneration. An attrition-driven reduction in female headcount concentrated in admin roles is a reportable outcome, not just an operational decision.

2. Redesign roles before cutting them

The cheapest option is attrition: skip the backfill, let AI absorb the tasks. The more defensible option, operationally and for WGEA compliance purposes, is to rebuild the role around what AI cannot yet do consistently: exception handling, stakeholder coordination, and judgement calls on ambiguous cases. A rebuilt administrative role often sits in the same classification band under the relevant Modern Award (for example, the Clerks Private Sector Award 2020 covers many of these positions), but generates more organisational value and retains the tacit knowledge that experienced admin staff hold about how the business actually functions.

3. Fund reskilling in the direction of the exposure gradient

Generic AI literacy training, a link on the intranet and a sixty-minute module, does not move the needle on retention or capability. What does move it: paying for a bookkeeper to sit the Certified Practising Accountant (CPA) or Chartered Accountants ANZ foundation exams, funding a payroll clerk's transition into HR business partnering, or moving a receptionist into a client-success role where interpersonal skills become a genuine differentiator rather than a baseline job requirement. The reskilling investment has to point toward roles that are actually growing in the current vacancy data, and it has to be funded by the employer, not offered as a self-directed option the employee is expected to pursue in their own time.

The bigger picture

For most of the last decade, the public debate about automation risk centred on factory floors and long-haul transport, a story weighted toward men in physical occupations. The JSA exposure data reframes that. The occupations losing job openings fastest right now are the ones where women sit at desks. The occupations still growing are the ones where workers wear high-vis and carry a trade licence. If that gradient is left to run without deliberate intervention, the gender employment gap widens by default, not by design. The Australian Bureau of Statistics (ABS) Labour Force Survey already shows women are more likely than men to work part-time in clerical roles, which amplifies the income impact of any contraction in hours or headcount in this segment.

For an individual worker, the useful frame is that your job title matters less than the mix of tasks inside it. For an HR team, the useful frame is that the vacancies you are not posting this quarter are the pipeline decision, regardless of whether anyone has written it down as policy. Both groups are working the same structural problem from opposite ends. The gap between those two vantage points is where company practice, and government policy through bodies like the Fair Work Commission (FWC) and the Office of AI, will determine who navigates this shift well and who does not.