AI is smart but still sexist? UN report shows it keeps getting women wrong

At least 44% of 133 AI systems, examined as part of a study, displayed gender bias, while 26% exhibited both gender and racial bias, according to UN Women. Despite these findings, only around half of marketers currently rely on human review to assess AI-generated content before it is published.

With artificial intelligence becoming an integral part of everyday life, UN Women has warned that the technology is perpetuating deeply rooted gender stereotypes. This is contributing to rising levels of online abuse and sidelining women from key discussions about the future of digital innovation, it stated.

The warning follows a series of recent studies indicating that while generative AI has become a routine part of daily life for billions of people, it is also reinforcing existing inequalities through biased algorithms.
Ahead of the United Nations Global Dialogue on Artificial Intelligence Governance, scheduled for July 6–7, and the AI for Good Global Summit in Geneva from July 7–10, UN Women has outlined the urgent reforms needed to ensure the digital future is more inclusive and equitable.

“As AI tools become embedded in content generation and media buying at scale, decisions about who gets seen, how they are portrayed, and whose stories get told are being made at speed, and largely without human scrutiny or gender perspective,” UN Women said in a press release.

Artificial intelligence systems powered by large language models continue to reproduce traditional gender stereotypes, according to research. Women were repeatedly associated with the home, family and children, while men were linked to professional success, leadership roles and higher earnings.

The study also found that approximately one-fifth of AI-generated sentence completions based on gender included sexist or misogynistic narratives, with some portraying women as sexual objects or the property of their spouses. These biases are rooted in decades of unequal representation within the data used to train AI.

UN Women cautions that inadequate policymaking could worsen the problem, noting that only 24 of 138 countries examined mention gender in their national AI strategies, and only 18 have adopted meaningful gender-responsive policies.

While generative AI is widely expected to fuel job creation in technology-focused sectors, women’s participation in STEM and AI remains disproportionately low. Women make up only 30% of the global AI workforce, leaving the development of these systems in the hands of teams that fall short of representing the billions of users who will ultimately rely on them.

Calling for a more inclusive approach to AI, UN Women said the rights, perspectives and lived experiences of women and girls should be reflected at every stage of AI development, implementation and governance. The organisation noted that AI can become a powerful tool for tackling stereotypes, expanding representation and improving accessibility if it is designed with appropriate safeguards.

It stressed that success will ultimately depend on policymakers, businesses and researchers ensuring that women and civil society groups have a meaningful role in shaping the technology.