UN Women urges governments and developers to embed gender equality into AI
The organisation has drawn on research from several studies to highlight the scale of gender bias in artificial intelligence systems and has set out what must change to build a gender-equal digital future
UN Women has called on governments and those developing artificial intelligence to embed gender equality and the rights and experiences of women and girls into “every stage of the AI lifecycle from development, deployment, and governance”.
The organisation highlighted research from the Berkeley Haas Center for Equity, Gender and Leadership that found that of 133 AI systems analysed, 44% demonstrated gender bias and 26% demonstrated both gender and racial bias.
In an article published ahead of the United Nations Global Dialogue on Artificial Intelligence Governance and the AI for Good Global Summit – which will take place in Geneva, Switzerland, between 6 and 10 July – UN Women set out “what is at stake – and what must change – to build a gender-equal digital future”.
“When designed with safety and used with intention, AI can help detect stereotypes, broaden representation, and improve accessibility at scale,” UN Women said.
“The choice of whether it does lies with the people making decisions – in governments, in companies, in experts researching and developing AI – and it depends on whether we incorporate the voice, expertise, and lived experience of women and girls from diverse contexts [and the] civil society organisations who work with them and know their issues deeply.”
Join the Global Government Women’s Network
LLMs exhibit sexist and misogynistic attitudes – and it’s a ‘policy problem’
The article noted that large Language Models (LLMs) have been found to consistently associate women with ‘home’, ‘family’, and ‘children’, and men with ‘business’, ‘executive’, ‘salary’, and ‘career’.
Research by UNESCO, for example, found that when tasked with completing sentences that start with a person’s gender, “about 20% of responses from LLMs exhibited sexist and misogynistic attitudes, including portrayals of women as sex objects and property of their husbands”.
It said “these are the predictable output of AI systems trained on decades of unequal representation of women and men”, and emphasised that AI bias is not only a system design problem, “but also a policy problem”.
Of 138 countries assessed in a UN Women study published in 2025, only 24 referenced gender in a national AI strategy, and just 18 included substantive gender-responsive provisions, which it said “risked inequality being ‘baked in’ to future systems”.
AI ‘intensifying violence against women and girls in digital spaces’
Data collected from a UN Women survey also showed that almost one in four women human rights defenders, activists and journalists had experienced AI-assisted online violence and that 12% report having experienced the non-consensual sharing of personal images, including intimate or sexual content.
Of the respondents, 6% said they had been targeted through deepfakes or manipulated images or video, while more than one in four had received unsolicited sexual advances through digital messaging.
“AI is compounding this. Deepfakes are among the most visible examples of AI-enabled abuse that disproportionately targets women and girls. As AI-generated content becomes the norm, the tools for harassment, manipulation, and image-based abuse are scaling alongside it,” UN Women said.
Read more:
Governments take action against Grok AI tool over sexualised images
UK government launches deepfake detection initiative as ‘urgent national priority’
Women being ‘locked out of rooms’ where AI is built
The article highlighted that generative AI is expected to drive job growth in tech-intensive sectors but that “women remain underrepresented in Science, Technology, Engineering and Mathematics (STEM) and AI, making up only 30% of the AI workforce globally”, according to the International Labour Organization.
“The people designing these systems are not representative of the billions of people the systems are expected to serve – and that glaring gap is compounding the problem,” it said.
The economic disruption caused by AI is also expected to “fall hardest one women”, with women outside the AI sector nearly twice as likely as men to hold jobs at high risk of automation.
“AI disparity does not manifest in gender inequality alone – harms are multiplied across race, disability, socioeconomic status, and geography,” it stressed. “The communities already most underrepresented in media and labour markets face the greatest risk of being left further behind.”
AI-assisted marketing and communications perpetuating discrimination
Another concern UN Women covered in its article is the use of discriminatory algorithms in marketing and communications, with generative AI “now among the most widely used technologies in day-to-day marketing and communications work”.
It cited research by the Institute of Practitioners in Advertising which found that in the UK alone, 88% of advertising and media agencies are using generative AI in some form, and separate research that showed that only 51% of marketers currently use human oversight to test AI-generated creative before release.
“Discriminatory algorithms could further perpetuate gender inequality and discrimination,” UN Women said. “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.”
Inclusive AI is not only a gender equality issue but a “commercial imperative”, the organisation said.
It pointed to a study by the Unstereotype Alliance – an industry-led initiative convened by UN Women – which it said “proved that inclusive advertising has a positive impact on business profit, sales and brand value”.
Brands that create inclusive advertising that is free of gender stereotypes were found to enjoy a +3.46% short-term sales and +16.26% long-term sales uplift; to be 62% more likely to be a consumer’s first choice; to have 54% higher pricing power; and to experience 15% higher customer loyalty.
“As AI becomes central to how campaigns are planned and produced, the brands that embed inclusion into those processes stand to gain – and those that do not, face significant reputational and commercial risk,” it said.
Read more: Estonia set to be first country to create digital identities for AI agents
About Mia Hunt
Mia has been editor of globalgovernmentforum.com since 2019. She has 15 years’ experience as a journalist and editor and specialises in writing for civil and public servants worldwide, including covering sustainability policy and related issues. She has led the Global Government Women’s Network since it launched in 2023. Previously, she covered commercial property having been market reports and supplements editor at Property Week and deputy editor at Retail Destination. She graduated from Kingston University London with a first-class honours degree in journalism and was part of the team that produced The River newspaper, which won Publication of the Year at the Guardian Student Media Awards in 2010.
Related Posts
