This report analyzes a scientific corpus consisting of 7,115 news stories, 13,787 journalistic sources, and a total of 11,266 photographs and images published between 2017–2021 and in 2024. All data was manually gathered by about twenty researchers who are currently developing, as part of another scientific project, an algorithm to automatically analyze biases in media coverage.
According to the report, the distribution of media focus regarding individuals appearing in the analyzed period is highly unequal by gender: 41.60% focuses on men; 39.06% is mixed; 14.01% adopts a neutral focus; and 5.33% focuses on women. The study also examines proper names in headlines: 13.11% of total news stories feature a man's name, while a woman's name appears in only 4.19% (the majority of headlines are neutral—entities—or contain no proper names). "The exclusive presence of women records lower visibility throughout the study, confirming a profound invisibility of women," the report concludes.
Regarding journalistic sources, men account for 61.14% of quotes compared to 21.45% for women, reaching a disparity of more than 4 to 1 in C-suite or presidential positions. "Men are interviewed far more often as experts than women," concludes the study's lead author, Clara Sainz de Baranda, Professor of Journalism in UC3M's Department of Communication.
Stories bylined by female journalists tend to include a higher proportion of female sources (25.58%) than those bylined by men (18.96%). Nevertheless, male bylines widely dominate newsrooms (49.02% vs. 27.51% for women). Furthermore, the massive surge of corporate and institutional sources in 2024 (29.90%) has tended to displace space for independent female experts.
Narratives regarding technical development, corporate strategy, and the economics of AI are almost entirely dominated by male or institutional voices and images. The space granted to women is concentrated in academic research (where they represent 29.26% of sources in that sector) and in analyzing the social, ethical, educational, or cultural impact of technology.
Sexist language and gender biases
The report also analyzes sexist language, showing that while it has improved overall in news media, biases remain, such as the use of the generic masculine. "When reading a news story that mentions 'experts,' the mind defaults to viewing experts as men. Therefore, female researchers and women in general must be made more visible in language," notes Clara Sainz de Baranda. Another bias highlighted is the semantic leap, which occurs when discourse appears inclusive of everyone but later reveals women were not actually included: "They speak of 'experts,' and then it turns out the award winners were women," she points out.
According to the report, among 2024 news items featuring women, 83.12% include some form of sexist language and 81.06% rely on the generic masculine. When covering female professionals, invisibility biases persist, such as referring to them by job title without their proper name (14.45% of cases), asymmetrical treatment (first name for the woman vs. surname for the man), or framing their success as an exceptional or anomalous event.
In journalistic iconography, men are portrayed in active roles of leadership and control over technology (32.41% of images vs. 14.26% featuring women). In representations of women (and in female robot or gynoid depictions), dynamics of objectification and sexualization reemerge (8.59% of photos of women in 2024 exhibit sexualization biases). The report notes that 37.13% of AI-generated illustrations or creations examined in 2024 reproduce traditional gender stereotypes.
Recommendations for journalists
Beyond analyzing these inequalities, the project includes a guide to good journalistic practices to promote more egalitarian AI coverage, developed by Clara Sainz de Baranda alongside Carmen María Navarro López of UC3M's University Institute of Gender Studies.
Among other recommendations, they propose diversifying sources through gender-balanced contact lists that give visibility to women in executive positions and technical development, abandoning the generic masculine and narratives of exceptionality when covering female talent, eliminating the objectification and sexualization of robotic depictions, encouraging sex-disaggregated data in news stories, and increasing female journalists in business and technology desks.
Among the tools suggested for reviewing content is the reversal rule, which consists of swapping female and male gender terms to test whether the change alters sentence perception or reveals potential stereotypes. "Applying this test can help identify unequal treatment and revise language to avoid sexist expressions," they note.
Bibliographic references:
Sainz de Baranda Andujar, Clara et al. (2026). InfoIA. Análisis de las informaciones sobre Inteligencia Artificial publicadas en medios digitales. Investigaciones Feministas. Instituto de las Mujeres O. A., Ministerio de Igualdad. e-Archivo de la UC3M: https://hdl.handle.net/10016/51005
Sainz de Baranda Andújar, Clara y Navarro López, Carmen María. (2026). IA e Igualdad. Guía de Buenas Prácticas para la Información Periodística. Investigaciones Feministas. Instituto de las Mujeres O. A., Ministerio de Igualdad. e-Archivo de la UC3M: https://hdl.handle.net/10016/51007
Video: https://youtu.be/sW4i0988Tas