Loading page…
The InfoIA research finds disparities in sources, language and imagery, with women’s visibility also varying by who writes the story.
Listen to this story
Universidad Carlos III de Madrid’s InfoIA report, publicized on October 22, finds that Spanish digital coverage gives men more prominence and authority in stories about AI, while women’s inclusion often comes with unequal language. The analysis covers 7,115 stories from 2017–2021 and 2024, not a continuous period through 2026. Its accompanying journalism guide turns the findings into editorial practices, including balanced expert lists and a test for language that treats women as exceptions.
Men were the focus of 41.60% of stories, compared with 5.33% focused on women; mixed and neutral coverage made up the remainder.
Men accounted for 61.14% of quoted sources and women 21.45%; the gap among top executives and presidents exceeded four to one.
In 2024 coverage featuring women, researchers classified 83.12% of items as containing sexist language; 81.06% used the generic masculine.
Appearing in an AI news story does not guarantee equal treatment. A newly publicized study from Universidad Carlos III de Madrid finds women far less visible than men in Spanish AI coverage. Even stories featuring women frequently used language the researchers classified as sexist, while quoted expertise leaned heavily toward men.
The university described the findings on October 22 in a research release on Newswise. Its InfoIA report, conducted by UC3M’s University Institute of Gender Studies, examined 7,115 news stories, 13,787 journalistic sources and 11,266 photographs and images. About 20 researchers manually collected the material, covering Spanish digital media in 2017–2021 and 2024—not a continuous period through 2026.
Men were the focus of 41.60% of stories, compared with 5.33% focused on women. The remaining coverage had either a mixed focus, at 39.06%, or a neutral focus, at 14.01%. Headlines showed a similar imbalance: men’s names appeared in 13.11% of all stories’ headlines, versus 4.19% for women. Most headlines named entities or contained no proper names.
The disparity also affected whose views readers encountered. Men accounted for 61.14% of quoted sources, compared with 21.45% for women. Among sources in top executive or presidential positions, the gap exceeded four to one. Lead author Clara Sainz de Baranda, a UC3M journalism professor, said men were interviewed as experts far more often than women.
The report describes technical development, corporate strategy and AI economics as dominated by male or institutional voices and images. Women’s presence was concentrated in academic research and discussion of technology’s social, ethical, educational and cultural effects. Within academic research sources, women represented 29.26%.
Source selection differed with the writer’s byline, too. Stories by female journalists included a higher proportion of female sources than those by male journalists. Men held 49.02% of bylines in the analyzed coverage, compared with 27.51% for women.
Women represented 25.58% of sources in stories by female journalists.
Women represented 18.96% of sources in stories by male journalists.
Among 2024 news items featuring women, the researchers classified 83.12% as containing some form of sexist language. They found 81.06% used the generic masculine: masculine wording intended to refer to people generally. Sainz de Baranda argues that this language encourages readers to picture experts as men.
Other patterns were more specific to how female professionals were introduced. The report identified references by job title without a proper name, using a woman’s first name while giving a man’s surname, and presenting women’s success as exceptional or anomalous. These findings concern treatment within coverage, not simply the number of women included.
The visual analysis extended to AI-generated material. The report found traditional gender stereotypes in 37.13% of the AI-generated illustrations or creations examined from 2024. It also identified sexualization biases in 8.59% of photographs of women that year.
The project includes a journalism guide developed by Sainz de Baranda and Carmen María Navarro López. Its recommendations, also detailed in the university’s account on Phys.org, address both reporting habits and editing choices:
Loading discussion...
Join the conversation
Explain where deliberate balance helps—or where you would draw a line.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.