Gender bias concerns in artificial intelligence systems

Research and reports from organizations including UNESCO, the Council of Europe, and the EU Agency for Fundamental Rights highlight systemic gender bias in AI technology. In healthcare, AI models often rely on male-centric data, leading to misdiagnosis or underestimated symptoms in women, a trend also noted by a 2025 LSE study in English municipalities. Facial recognition systems have been found to show higher error rates for women, particularly those with darker skin, compared to white men, as evidenced by studies from MIT, Stanford, and the EU. Furthermore, a 2024 UNESCO report analyzing 133 AI systems found that 44% exhibit gender bias by associating women with domestic roles and men with career-oriented terms. Recruitment algorithms have historically favored male applicants due to training data consisting primarily of male resumes, such as instances observed at Amazon and in general 2024 evaluations. Financial bias was also identified in 2019 regarding the Apple Card algorithm, which offered lower credit limits to women despite similar or better credit scores than men. Additionally, the European Parliament has identified concerns regarding the potential for AI to facilitate gender-based violence, specifically through deepfake technology.