The EU AI Act and recruitment when the public sector employer has to assess the fundamental rights risk itself
The Commission has opened a consultation on the draft template for the fundamental rights impact assessment. Here is who must use it from December 2027 and what it asks when AI is used in recruitment.
What happened
On 16 November 2026 the European Commission opened a consultation on a draft template for the fundamental rights impact assessment under the AI Act. The template belongs to Article 27, which requires certain deployers of high-risk AI systems to assess the impact of their use on fundamental rights before the system is put into use. According to the Commission's news item on digital-strategy.ec.europa.eu, the draft was prepared by the AI Office with input from the EU Agency for Fundamental Rights.
There are two deadlines. Deployers who want to test the template on one of their own systems should contact the AI Office by 4 December 2026. Public feedback can be submitted until 15 January 2027. The obligation itself applies from 2 December 2027. Once adopted, the template becomes an automated tool on the AI Act Single Information Platform, which deployers use to notify their results to the market surveillance authority.
Who has to carry out the assessment
The explanatory notice published with the draft makes clear that not every deployer of high-risk AI has to produce the assessment. The duty applies to bodies governed by public law, meaning government bodies and public institutions, and to private entities providing public services, when they use high-risk AI in one of the areas listed in Annex III. It also applies to deployers of AI used for credit scoring and for risk assessment and pricing in life and health insurance.
Employment, workers' management and access to self-employment is one of the Annex III areas and appears as a tick box in the template's first section. A municipality, a hospital or a university using an AI-based screening or assessment tool in recruitment would therefore be covered. The notice names education, healthcare, social services, housing and the administration of justice as examples of public services private entities may provide, and says the list is not exhaustive. A private company recruiting for its own operations is not among the deployers the notice says must carry out the assessment.
Public authorities using high-risk AI in employment, among other areas, must also register a summary of their findings in the EU's public database of high-risk AI systems. For a public sector employer, the assessment of a recruitment tool is therefore not an internal document.
What the template asks
The template follows Article 27(1) and has seven sections. After the general information in section one, the deployer describes the intended purpose of the system as defined by the provider in the instructions for use, and whether its own use stays within that purpose. Then come the processes the system will be used in, how it produces its outputs, how those are used in practice, and for what period and how often.
The third section concerns the groups likely to be affected. The template lists, among others, children, older people, women, persons with disabilities, persons in a vulnerable social or economic situation and persons with a minority racial or ethnic background. In the fourth section the deployer describes the specific risks of harm, rates likelihood and severity on a three-point scale and ticks the rights at stake, which include equality and non-discrimination, protection of personal data, freedom to choose an occupation and workers' rights. An existing data protection impact assessment may be cross-referenced.
The fifth section requires a concrete description of human oversight, meaning who carries it out, what authority that person has to override, disregard or reverse the system's output, what tools and procedures support them, and how the deployer ensures they have the necessary competence, training and authority. The sixth covers what happens if risks materialise, including monitoring, suspending the system and handling complaints from affected individuals. The seventh is an overall conclusion.
What it means for assessment work
For a public sector employer using or considering an AI-based assessment tool, the template moves a number of questions from the vendor to the deployer itself. The assessment should rest on the provider's instructions for use and technical documentation, obtained, reviewed and kept available before the template is completed. The notice also recommends a multidisciplinary team combining fundamental rights expertise, technical understanding and knowledge of the recruitment process.
At the same time, a deployer may rely on a previous assessment or on one carried out by the provider, where the system and the identified risks are similar. The provider is not obliged to share its assessment, but the notice calls sharing good practice. That makes it a relevant tender question whether the vendor can provide such an analysis, and whether the instructions for use describe the oversight measures the deployer has to put in place itself.
The draft has not been adopted and reflects the preliminary views of the Commission's services. But the structure is known now, and with the obligation applying from December 2027 there is about a year to work out whether the organisation is covered and, if so, which documentation is still missing.
Source: https://digital-strategy.ec.europa.eu/en/news/commission-seeks-feedback-draft-template-fundamental-rights-impact-assessment-under-ai-act