AI Recruitment Tools Shown to Systematically Disadvantage Women, Study Finds

A landmark study by researchers at University College London and the Oxford Internet Institute has found that AI recruitment screening tools used by major employers across the United Kingdom, United States, and Germany systematically produce lower scores for female candidates compared with equally qualified male candidates. The research analysed over 200,000 anonymised job applications processed by seven commercially available AI screening platforms over an eighteen-month period.

How the Bias Operates

Researchers found that the AI systems assigned lower scores to women in sectors including finance, engineering, and technology — despite identical academic qualifications, work histories, and skills. The algorithms, trained predominantly on historical hiring data, appeared to replicate and amplify patterns from industries where male candidates had historically been preferred. In some cases, the mere presence of words associated with women's experiences — such as mentions of career breaks or maternity leave — further reduced scores, even when such information was not supposed to influence hiring decisions.

Calls for Regulation

The study's authors are calling for mandatory algorithmic auditing of all AI recruitment tools used in the EU and UK, arguing current self-regulatory frameworks are insufficient. Several companies that use the tools said they take algorithmic fairness seriously and were reviewing their systems in light of the findings. The EU AI Act, which partially came into force in 2025, classifies AI hiring tools as high-risk systems requiring transparency and human oversight, but critics say enforcement has been too slow.