On 7 October, LSTS researcher Anastasia Karagianni gave a guest lecture in the course Anti-discrimination Law: European and Comparative Perspectives and a presentation at an internal seminar of the Law & Tech Working Group at Science Po Paris.
The lecture, entitled “Gendered Algorithms and the EU AI Act: Feminist Approaches,” examines how gender-based discrimination emerges in AI systems in practice, drawing on case studies of biased automated decision-making.
The visit takes place at the invitation of Raphaële Xenidis.
Anastasia also presented ongoing work during an internal seminar of the Law & Tech Working Group entitled “Verifemmication in Action: An Intersectional Feminist Framework for Evaluating Vulnerability in AI Systems.”
Abstract
The talk introduces verifemmication, a socio-legal evaluative framework for analysing AI systems through the lenses of gender equality, intersectionality, and epistemic injustice. It addresses a key limitation in the AI Act, where compliance-oriented tools, such as fairness metrics, risk assessments, and harmonised standards, often fail to capture structural inequalities, relational harms, and forms of epistemic exclusion embedded in AI systems. Instead of focusing on whether an AI system meets compliance criteria, the framework shifts attention to deeper evaluative questions, like whose knowledge defines what is considered risk, whose safety is prioritised, and at what cost, which harms remain invisible within dominant evaluation practices. The discussion is grounded in feminist epistemologies and Science and Technology Studies (STS).