Source: Statewatch

A letter from the staff committee of the Court of Justice of the European Union dated 2 July demands that the European Commission explain plans to use an AI tool to screen candidates for permanent administrator roles, after the committee says it only learned of the project through press reports. Over 170,000 people applied this year for entrance exams while only 1,490 vacancies exist, and the Commission has launched a pilot to complement EPSO and institution-led selection with an AI-based Job Matching Application developed by Accenture using Anthropic’s LLM and hosted on Amazon Web Services, which was revealed by Euractiv on 15 June and is scheduled for summer testing.

The staff committee raises legal and procedural concerns about compliance with equal treatment, non-discrimination, transparency and good administration, and demands strong personal data protections before deployment. The letter cites the AI Act’s classification of recruitment tools as high-risk and warns that algorithmic screening and recommendations can be “decisive” even where humans make final decisions; it also references a Privacy International investigation published on 9 July about opaque and unreliable AI recruitment platforms. The committee questions the Commission’s assurances that candidate data will not be used to train models, warns of threats to EU digital sovereignty from outsourcing to US firms, flags a potential conflict with the Court’s duty of impartiality if it may later judge AI recruitment cases, and calls for greater scrutiny to ensure respect for EU data protection, fundamental rights and the AI Act.