Home / Blog / Gender Equality in AI

AI Still Files Us Under "Home and Family"

Two new June reports on what AI gets wrong about women — and what to do with that information.

Key takeaway: Two new studies dropped this month. One found 44% of AI systems show gender bias. The other found women still use AI less — and trust it less. Here's why that matters for you.

By Dear Sarah · 2026-06-27 · Updated 2026-06-27

A Black woman in business attire focused on her smartphone, captured as part of the Women of Color in Tech series

There's a line from a UN Women report this month that I keep coming back to. Jayathma Wickramanayake, who leads digital technologies for UN Women, said AI models "pull bias from decades of text written by people, about people, in a world where women were filed under home and family."

Filed under home and family. That phrase did something to me.

What just came out

Two reports landed in the same week and they belong on the same page.

On June 22, the UN published a sweeping look at gender bias in AI. Of 133 systems studied, 44% showed gender bias, and more than a quarter showed both gender and racial bias. When researchers asked large language models to complete sentences about gender, roughly one in five responses came back sexist or misogynistic. Only 24 of 138 countries even mention gender in their national AI strategies.

Five days earlier, on June 17, Pew Research — led by women including Monica Anderson, Michelle Faverio, Eugenie Park, and Colleen McClain — published its 2026 AI gender gap study. The headline was that the gap in chatbot use has basically closed (50% of men, 47% of women). But the texture underneath is more interesting. Men use AI daily at higher rates (27% vs. 20%). Women are far more likely to feel AI is moving too fast (68% vs. 58%). And women are about twice as likely to expect AI's impact on their lives to be negative.

Why this lands for you

The instinct, when you read "AI is biased against women," is to either tune out or feel small. Don't do either.

What the data is actually saying is this: the tools were trained on a world that didn't take you seriously, and they will reflect that back unless somebody pushes. The women who built the Pew study are pushing. Joy Buolamwini and the Algorithmic Justice League have been pushing for years. Wickramanayake is pushing from inside the UN. You can push too, just by refusing to treat any chatbot's first answer as the truth about you.

If you're hesitating to use these tools because they feel off, you're not paranoid — you're noticing something real. The fix isn't to opt out. The fix is to use them while you stay the editor of your own story.

One thing to try this week

Pick a chatbot you already use. Ask it for advice on something that matters to you — a salary range for your role, a career pivot, how to handle a hard conversation with someone you love. Then ask it the same question reframed as if you were a man with the same circumstances. Notice what shifts. Notice what assumptions show up only for her.

You're not doing a research study. You're learning what your tools quietly think about women like you. Information you can use.

Quote to sit with

"Whoever codes the system embeds her views. Limited views create limited systems." — Joy Buolamwini

💌 Sarah

Whoever codes the system embeds her views. Limited views create limited systems. — Joy Buolamwini
  • #ai
  • #gender-equality
  • #bias
  • #research
  • #agency

Sources

  • AI is getting women wrong as gender bias persists, data reveals — UN News
  • AI is getting women wrong as gender bias persists, data reveals — UN Geneva
  • The gender gap in AI in the US — Pew Research Center
  • InCoding — In The Beginning Was The Coded Gaze — MIT Media Lab