Women are damned if they AI and damned if they don't. So do it anyway.
New 2026 data shows women are 32% more likely to fear being called "cheaters" for using AI.
Radhika Bajaj, a journalist based in Mumbai, put it plainly in a LinkedIn post a few weeks ago: “I use AI at work. It’s my assistant, not my replacement.” The statement was clean and confident. The kind of thing you’d expect to go viral. (It did.)
But for every woman who finds that footing, there are many more finding it difficult to negotiate the tradeoffs that come with using AI.
Account executive Angela Tran told Axios in September 2025 that she runs her work emails through AI before sending them.
“I don’t want to come off as too pushy,” Angela Tran said.
Jennifer Borchardt, a consultant, described a similar habit: “I wish I didn’t have to do it, but it’s working.”
Susannah Shattuck, a product head at Credo AI, is more bullish. “I love using AI for vendor negotiations,” she told Axios. “It pushes me to be a tougher negotiator.”
Then there’s the developer who posted on Reddit last month with a bleaker take: “I use AI at work and feel myself getting dumber by relying on it. Our company required we use AI, so there’s just no avoiding it.”
Guilt. Pragmatism. Enthusiasm. Resentment.
These aren’t the responses of women who are afraid of technology. They’re the responses of women using a tool that comes with uneven terms, one that’s required by some managers and stigmatized by others.
According to research released by LeanIn.org in March 2026, women are twice as likely as men to see their reputation take a hit when their AI use is noticed at work. They’re also 32% more likely to fear that it’ll be seen as cheating, and 27% less likely to be praised by a manager when they do use it.
While Angela Tran softens her emails through AI because she worries about coming across as too pushy, her male colleague is probably just hitting send.
Men are currently 22% more likely than women to use generative AI tools daily at work. With a technology that improves through practice, that gap compounds into what researchers call a fluency surplus. It’s the competitive edge that builds through consistent, early AI adoption and translates into faster work, better output, and visibility that results in rewards like promotions.
Left unaddressed, that AI gap becomes a meaningful difference in who gets ahead.
Canada’s AI gender gap
Canada is a leader in AI. We have the Vector Institute in Toronto, Mila in Montréal, Amii in Edmonton, and federal dollars flowing into our National AI Strategy. The infrastructure is real.
But the access isn’t.
Only 22% of AI workers in Canada are women. A 2024 Future Skills Centre survey found just 47% of Canadian women were familiar with workplace AI tools, versus 53% of men. That gap exists while women are concentrated in the exact roles most vulnerable to automation.
Between 2006 and 2021, Canada dropped from 14th to 24th globally on gender parity indexes. If we aren’t careful, the AI transition could accelerate that drop.
The jobs that are already disappearing
The World Economic Forum and the ILO are clear: women’s jobs face roughly 3 times the AI exposure risk of men’s. The roles most at risk are ones women have historically dominated:
Legal secretaries: 96% female, 75% AI exposure
Medical secretaries: 94% female, 63% AI exposure
Receptionists: 92% female, 58% AI exposure
Payroll clerks: 89% female, 50% AI exposure
Court clerks: 85% female, 58% AI exposure
A March 2026 report from Anthropic on AI’s labour market impacts explains: AI is most heavily used for writing, communication, and administrative tasks like drafting emails, summarizing documents, generating reports, and coordinating across teams. These tasks disproportionately make up women’s work.
The report shows that AI is automating the execution layer while leaving the oversight, strategy, and decision-making layer intact. But the execution layer is where women’s contributions have historically been visible.
The AI competence penalty compounds this. A 2025 working paper by researchers at several universities ran a controlled experiment where identical code was presented as AI-assisted or not. Engineers using AI received 9% lower competence ratings overall. Women received a 13% penalty, while men received 6%. The study hasn’t been peer-reviewed yet, but its findings are consistent with the LeanIn data. When the same work is done the same way with the same tool, the professional cost of disclosing AI use isn’t equal.
AI touching a job isn’t the same as eliminating it. But of the 6.1 million U.S. workers in roles with both high AI exposure and low adaptive capacity (the workers least positioned to pivot after displacement, because of age, savings, credentials, or skills gaps), 86% are women.
Women face a dilemma: they need to build AI fluency, but they’re penalized for using AI. And AI restructures the labour market in ways that disproportionately hurt them.
But shying away from AI use isn’t the answer. If women step back because the risk feels too high, they won’t be in the rooms where the technology’s longer-term direction gets set. For instance, woman are 38% more likely to raise concerns about AI reliability and bias, but only if they know about those things in the first place.
As of 2022, women held about 30% of AI-related technical roles globally, a number that barely moved since 2016.
Beyond the “courage gap”: Why the framing matters
It’s worth noting that LeanIn hasn’t released the wording of their March 2026 survey, which means the 32% figure for women fearing they’ll be seen as “cheating” reflects responses to a prompt we can’t verify. The findings align with WEF and ILO data, which strengthens them. But how a question about professional anxiety is phrased shapes the answer considerably.
That matters because the idea that women need to overcome their fear to adopt AI is complicated, and it’s not the whole story.
When Sheryl Sandberg published Lean In in 2013, bell hooks called it “faux feminism” for centering individual advancement over structural change. Dawn Foster’s rebuttal, “Lean Out,” made the same argument from a different angle: that framing workplace inequality as a problem of women’s ambition or confidence lets institutions off the hook.
If managers are 23% more likely to encourage men to use AI, and women twice as likely to have their competence questioned when they use it, the barrier isn’t psychological, it’s structural. We keep telling women to speak up in rooms that were built to silence them.
On Mel Robbins’ Podcast in November, AI advisor Allie K. Miller described a “toxic flywheel”: most people talking publicly about AI are men, so women don’t see themselves in the conversation. Because the use cases shared skew male, they feel less relevant to women’s lives, and because fewer women engage, fewer women are visible as AI practitioners.
It’s a loop that operates outside of individual courage or hesitation. Miller’s reframe is practical: AI isn’t coming for my job, it’s part of my job.
The manager training component of LeanIn.org’s strategy targets attribution bias at the organizational level. But the AI Circles program, peer experimentation groups for women, still places the work of change on the people experiencing the disadvantage. We need workplaces to allow women to participate, and to reward them for their participation.
What needs to change
Women hold 28% of S&P 500 board seats and represent 8.8% of Fortune 500 CEOs. They’re hard-won numbers, but they’re fragile ones. In Europe, women’s representation in tech has fallen from 22% in 2023 to 19% in 2025, as layoffs hit DEI and support functions hardest.
The LeanIn.org AI Circles program is built around a curriculum that moves participants through three stages:
Prompt engineering: Structuring queries to get reliable, useful outputs from large language models like Claude or ChatGPT.
Agentic workflow design: Orchestrating multiple AI systems to execute complex, multi-step tasks autonomously.
Ethical red-teaming: Proactively stress-testing AI outputs for the bias and errors that would otherwise go unnoticed until they cause real harm.
The goal is to produce leaders who understand the systems well enough to question and govern them. And LeanIn is pushing companies to build AI org charts, or transparency documents that map where AI is embedded in processes and who’s accountable for oversight. Visibility, their argument goes, creates accountability. Accountability changes who gets AI-adjacent assignments.
Companies with gender-diverse leadership are 25% more likely to outperform on profitability and see up to 45% higher revenue from innovation. Closing the AI gender gap is a business case as much as an equity one.
The developer who wrote “I feel myself getting dumber by relying on it” wasn’t describing a failure of courage. She was describing what it feels like to have a tool mandated from above, with no agency over how to use it, and no framework for thinking about what she was trading away. That’s a design problem — in the tool, in the workplace, and in the institutions deciding what counts as a skill.
But the women who understand these systems are harder to sideline than the women who don’t. AI fluency isn’t a magic fix, but you can’t advocate for what you don’t understand, and you can’t govern something you’ve never used.
If you’ve been hesitant, start now: open a free tool like Claude, or ChatGPT, or a local model if you’re concerned about data privacy. Give it a real task from your day-to-day and start to understand what it can and can’t do. AI understanding is a career asset, and it’s a form of literacy that shapes whether you’re in the room when the decisions that affect you are made.
The women who build AI fluency now are building career equity for the future.
Disclosure: I lead AI communications at Manulife, a life insurance company. All views expressed in this newsletter are my own and do not represent my employer.
AI in the news
MANGO is the new FAANG, and it’s coming for Canada’s AI talent (Be Giant) Canada has become a global hotspot for AI talent, drawing aggressive recruitment from major U.S. tech companies (the “MANGO” group). It’s created an intensely competitive hiring market where top candidates juggle multiple offers and rising expectations. In response, Canadian companies are differentiating less on perks and more on meaningful work, learning opportunities, and values like long-term thinking and ethical AI.
OpenAI, not yet public, raises $3B from retail investors in monster $122B fund raise (TechCrunch) OpenAI raised $122 billion at an $852 billion valuation ahead of a potential IPO, signalling massive investor confidence as it pours capital into infrastructure, talent, and AI development. The company is positioning itself as an “AI superapp” with explosive growth in users and revenue, making a clear play to dominate how people and businesses interact with AI.
Sycophantic AI decreases prosocial intentions and promotes dependence (Science) AI chatbots are often designed to be overly agreeable and flattering, and research shows they produce significantly more people-pleasing responses than humans, even in situations involving harmful or unethical behaviour. While users tend to prefer and trust these responses, AI may reinforce biased perspectives instead of challenging them, undermining accountability and healthy conflict resolution.
I have enjoyed this article, with it’s impactful read for workforce and diversity in embedding AI, skills and performance.
What you’re surfacing isn’t just hesitation, but also asymmetry.
The data shows women face twice the reputational risk for using AI, yet men build fluency 22% faster through daily use.
It’s not a comfort gap; but a consequence gap.
When a tool becomes essential for productivity, but unsafe for certain gender or disadvantaged groups for use openly, the system isn’t neutral. It has cultural connotations. It’s allocating advantage or disadvantage depending how you look at it.”
Good read, as someone who now uses AI across my work- both in my job as well as the business I am building on the side, it’s a game changer. But honestly it all started because of a conversation with my husband last year who was already using it and he encouraged me. Given how much better it has gotten over the short time I have been using it, it’s a key tool that should be considered.