These questions were in focus of a recent online event organised by the Council of Europe. The webinar marked the launch of two new publications, “Legal protection against algorithmic discrimination in Europe: current frameworks and remaining gaps” and “European policy guidelines on AI and algorithm-driven discrimination for equality bodies and other national human rights structures” took place during the webinar.
The event and the publications were developed in the context of the project Upholding equality and non-discrimination by Equality bodies regarding the use of artificial intelligence (AI) in public administrations. The project is co-funded by the European Union through the Technical support instrument and implemented by the Council of Europe in co-operation with the European Commission.
Algorithmic bias worsens existing social inequalities
AI is increasingly permeating every aspect of our daily lives, presenting significant challenges to the protection of fundamental rights. Algorithmic discrimination emerges as a particularly pressing concern. Research has demonstrated that algorithmic bias not only reflects but also exacerbates existing social inequalities.
In employment, selection algorithms trained on historical hiring data could unfairly favour male candidates by reproducing past hiring decisions that were skewed by discriminatory stereotypes against women or minority groups. Some of the profiling tools used by public employment agencies to help decide on allocating resources to jobseekers, have been found discriminatory. AI and automated decision-making (ADM) systems are regularly used by law enforcement authorities, but research shows that face recognition technologies can exhibit discriminatory biases or be deployed in a manner disproportionately targeting and monitoring minorities and thus carrying out ethnic profiling. AI and ADM systems are also used by certain public administrations in Europe to support decision making in the field of migration, for example regarding decisions on citizenship, asylum or residence. Reported uses include, for example, language identification and assessment, detecting fraud related to identity documents, case management, interacting with migrants including through chatbots, migration forecasting and border surveillance technologies. Other policy areas where public administrations are using AI and ADM are welfare, justice, education, tax, or healthcare.
In the private sector, personnel services companies use AI and ADM to screen and profile job applicants, match job applications to their profiles, or draft and translate job vacancies. Such systems are also deployed in such critical areas of the private sector, as banking and insurance.
New legal frameworks
In 2024 both the European Union and the Council of Europe adopted landmark legal instruments: the EU AI Act (Regulation 2024/1689) and the Council of Europe Framework Convention on artificial intelligence and human rights, democracy and the rule of law. One of the publications presented at the event today examines how these emerging legal instruments contribute to strengthening protections against algorithmic discrimination in Europe and assesses the lacunae that continue to affect these legal frameworks.
Helping actors at the national level
The second publication focuses on the key role that equality bodies and other national human-rights structures can play in identifying and mitigating the risks presented by AI/ADM systems and in promoting fundamental rights-compliant deployment of AI/ADM systems by public sector organisations. The guidelines offer recommendations and examples for applying new EU and Council of Europe regulations and are meant to assist and advise national stakeholders in the fields of human rights, equality and non-discrimination.
Representatives of the European Commission, national equality bodies from Belgium, Portugal and Finland, and academics were among the participants.
The Council of Europe runs extensive programmes of cooperation activities to help combat discrimination, hate speech and hate crime, as well as support the empowerment of minorities, the positive management of diversity and respect of human rights, including in the deployment of digital technologies, algorithmic and artificial-intelligence systems.
Read “European policy guidelines on AI and algorithm-driven discrimination for equality bodies and other national human rights structures” in full (also in FRA, DUT, POR, FIN, SWE)
Read “Legal protection against algorithmic discrimination in Europe: current frameworks and remaining gaps” in full (also in FRA, DUT, POR, FIN, SWE)
The Council of Europe and inclusion and anti-discrimination
Framework Convention on artificial intelligence and human rights, democracy and the rule of law