Last year, I wrote about the exciting potential of artificial intelligence (AI) for progressing gender equity in the workplace.
Almost a year on, the research paints a concerning picture: many employers are still struggling to harness that opportunity and, in the meantime, AI has demonstrated its ability to actively increase barriers and widening gender gaps in their organisations.
Here’s an overview of what the data shows:
- The roles most exposed to automation are disproportionately held by women, with the UN Women Gender Snapshot 2025 revealing that 9.6% of women’s jobs are at high risk of AI displacement, compared to just 3.5% of men’s.
- Women are already feeling the impact, with WORK180 research revealing that around one in five say AI is making their roles feel less secure, and a further one in five feel nervous using it at work.
- Of the women who feel nervous using AI, almost half (47%) cite a lack of clear rules or guidance revealing a concerning gap between workplaces’ rate of AI adoption and the support structures around it.
- What’s more, new technology enabled threats are also emerging that most workplace policies simply weren’t designed to address. For example, one in seven Australian workers has already experienced workplace sexual harassment facilitated by technology.
To be clear, none of these challenges are a result of women being unwilling or unable to engage with AI. Both WORK180 and the wider research point to the same conclusion: the barriers are mainly organisational, with a lack of clear guidance, inconsistent access to training, and AI decisions being made without the input of the women most affected by them.
The good news is that this means it’s a challenge that employers themselves have the power to solve.
Here are four actions employers can (and should) take right now
1. Build AI training and guidance that actually works for everyone
WORK180’s 2026 What Women Want Report revealed that many employers have moved fast on AI adoption but slow on the guidance, training, and support structures that make it accessible for all.
“We’re pushed by execs to implement AI but they don’t even know how and what tools to use so it’s overwhelming. Workload is heavy so they say automate as much as you can, and there’s not enough time to do the work and figure out how to use AI efficiently.”
WORK180 What Women Want respondent
💡What can you do? Audit what AI training and guidance currently exists in your organisation and who has actually accessed it. If the answer is patchy, inconsistent, or skewed toward certain roles, seniority levels, or demographics that’s where to start.
Communication needs to be considered here too. With 49% of What Women Want respondents citing concerns about the ethics or fairness of AI, it’s important to make sure your company has considered and clearly communicated this. What does use of AI mean at your business, and how does it align with your organisational values and behaviours?
2. Create psychological safety around AI use
Psychological safety ranked as the joint top priority for women assessing a new employer in our What Women Want 2026 research, with 70% of respondents rating it five out of five in importance.
This matters directly for AI adoption too. If the workplace culture doesn’t support speaking up, taking risks, or admitting uncertainty, it won’t support confident AI adoption either. This is true for anyone, but particularly for women and marginalised groups who can already face greater scrutiny at work.
💡 What can you do? Consider how AI use is encouraged and modeled across the organisation, including from leaders and managers. The key is to create space to learn, experiment and ask questions without fear of judgment.
3. Put your people at the centre of AI decisions — before and after deployment
Gender bias in AI doesn’t appear from nowhere. It’s inherited from historical data, and historical data reflects historical decisions. In most organisations, those decisions have favoured men, meaning a résumé-scoring tool or talent identification system trained on past data will quietly replicate that pattern, at scale.
The two moments that matter most are before a tool goes in and after it’s embedded.
Before: If the people most affected by an AI tool aren’t involved in selecting or implementing it, the tool is unlikely to solve the right problems and holds the potential to create new ones. Consider talent identification tools that surface “high potential” employees based on historical promotion data. If women were underrepresented in those past promotions, they’ll be underrepresented in the new AI outputs too.
After: By now, your teams are probably already using a variety of AI tools across the organisation. The question is whether anyone has checked them for gender bias, and whether your people have what they need to spot and address it. Bias can show up in the words a tool chooses, in what information it treats as important, and in who it surfaces and who it doesn’t. It isn’t always obvious, which is precisely why it needs to be actively tested for, not assumed away.
💡 What can you do? Map the AI decisions made in your organisation over the last 12 months. Who was in the room when those tools were selected? And has anyone reviewed the outputs for bias since?
For generative AI, prompts and workflows can be designed to actively flag bias. For tools used in recruitment, performance, and promotion decisions, build human oversight and bias-awareness as standard, not a one-off check.
4. Review your policies for new technology risks
As AI adoption grows, so does the range of technology-related risks employers need to plan for. One that’s largely missing from workplace policies is technology-facilitated abuse, where everyday tools like email, Teams, Slack, and location-sharing apps are used to monitor, harass, and coerce.
The Australian Government’s 2026 Status of Women Report Card identified this as a growing threat to women’s safety and workforce participation. Under Work Health and Safety laws and the Sex Discrimination Act 1984, employers have a legal positive duty to prevent it. It’s classified as a psychosocial hazard, sitting alongside your other workplace risk obligations.
Most employers have a sexual harassment policy. Far fewer have one that explicitly addresses what happens when technology is the vehicle for that harassment.
💡 What can you do? Review your existing policies against the eSafety Commissioner’s employer guidance on online abuse and Safe Work Australia’s online abuse resources. If technology-facilitated risks aren’t explicitly covered, that’s where to start.
The opportunity is still there
None of this means AI is a threat to gender equity by default. The opportunity I wrote about last year remains real; AI can take over repetitive tasks that fall disproportionately on women, free up time for higher-visibility work, and when governed well, reduce the role of subjective bias in key career decisions.
But that outcome requires deliberate choices. The organisations making the fastest progress on gender equity right now are not waiting until they have all the answers. They’re designing inclusive training, building diverse governance, checking their tools, and creating the conditions for everyone to engage with AI confidently.
The landscape will keep changing. The question is whether your organisation is shaping that change, or reacting to it.
