Skip to content

Knowledge base · AI in HR

May an AI tool infer an applicant's origin, religion or sexual orientation from photos or videos?

No. Since 2 February 2025 it has been prohibited to use a system that categorises people individually on the basis of biometric data in order to infer their race, political opinions, trade union membership, religious or philosophical beliefs, sex life or sexual orientation. Apart from that, such data is a special category of personal data and discrimination on those grounds is prohibited in recruitment.

In force now (since 2 February 2025)Trusq editorial · checked against the official textLast verified:

Next step: Map your AI tool

The prohibition

It is prohibited to use biometric categorisation systems that categorise individual persons on the basis of their biometric data to deduce or infer their race, political opinions, trade union membership, religious or philosophical beliefs, sex life or sexual orientation [1]. That prohibition has applied since 2 February 2025 [2].

What biometric categorisation is

An AI system for assigning people to categories on the basis of their biometric data [3], for example on the basis of facial features in a video interview.

Other characteristics

If a system categorises people on other sensitive or protected characteristics, then, to the extent that categorisation is not prohibited, it can be classified from 2 December 2027 as a high-risk system under Annex III [4] [5]. In addition, the GDPR and other applicable prohibitions remain in force [6].

Even without the AI Act

As a rule you may not process data revealing origin, religion or sexual orientation [6], and discrimination on those grounds is prohibited when offering a position [7]. Ask suppliers of video interview tools which characteristics their system infers from image or voice.

Sources

  1. [1] AI Act Art. 5(1)(g) (AI Act, Regulation (EU) 2024/1689). Prohibition of biometric categorisation to infer, among other things, race, religion or sexual orientation.
    (g) the placing on the market, the putting into service for this specific purpose, or the use of biometric categorisation systems that categorise individually natural persons based on their biometric data to deduce or infer their race, political opinions, trade union membership, religious or philosophical beliefs, sex life or sexual orientation;
    Source, retrieved on 3 October 2026.
  2. [2] Regulation (EU) 2026/1744 Art. 1(40)(a) (replaces AI Act Art. 113, third paragraph, point (a)) (Regulation (EU) 2026/1744 (Digital Omnibus on AI)). Chapters I and II of the AI Act (including the definitions, Article 4 and the prohibitions of Article 5) have applied since 2 February 2025.
    Chapters I and II shall apply from 2 February 2025, with the exception of Article 5(1), first subparagraph, points (ba) and (bb), and Article 5(1a) and (1b) which shall apply from 2 December 2026;
    Source, retrieved on 3 October 2026.
  3. [3] AI Act Art. 3(40) (AI Act, Regulation (EU) 2024/1689). Definition of a biometric categorisation system.
    ‘biometric categorisation system’ means an AI system for the purpose of assigning natural persons to specific categories on the basis of their biometric data,
    Source, retrieved on 3 October 2026.
  4. [4] AI Act, Annex III, point 1(b) (AI Act, Regulation (EU) 2024/1689). Biometric categorisation on the basis of sensitive or protected characteristics is listed in Annex III.
    (b) AI systems intended to be used for biometric categorisation, according to sensitive or protected attributes or characteristics based on the inference of those attributes or characteristics;
    Source, retrieved on 3 October 2026.
  5. [5] Regulation (EU) 2026/1744 Art. 1(40)(b) (replaces AI Act Art. 113, third paragraph, point (c)) (Regulation (EU) 2026/1744 (Digital Omnibus on AI)). The high-risk rules (Chapter III, Sections 1 to 3) apply to Annex III systems from 2 December 2027.
    Chapter III, Sections 1, 2, and 3, with the exception of Article 6(5), shall apply from: (i) 2 December 2027 as regards AI systems classified as high-risk pursuant to Article 6(2) and Annex III; and
    Source, retrieved on 3 October 2026.
  6. [6] GDPR Art. 9(1) (General Data Protection Regulation (GDPR), Regulation (EU) 2016/679). Processing of special categories of personal data is prohibited in principle.
    1. Processing of personal data revealing racial or ethnic origin, political opinions, religious or philosophical beliefs, or trade union membership, and the processing of genetic data, biometric data for the purpose of uniquely identifying a natural person, data concerning health or data concerning a natural person's sex life or sexual orientation shall be prohibited.
    Source, retrieved on 3 October 2026.
  7. [7] AWGB Art. 5(1)(a) (Dutch General Equal Treatment Act (AWGB); Dutch law, unofficial translation). Discrimination is prohibited when offering a position. Original Dutch text:
    Onderscheid is verboden bij: a. de aanbieding van een betrekking en de behandeling bij de vervulling van een openstaande betrekking; b arbeidsbemiddeling; c. het aangaan en het beëindigen van een arbeidsverhouding;
    Source, retrieved on 3 October 2026.

Related questions

May we use AI that infers the emotions of applicants or employees, for example in video interviews?May an AI selection tool distinguish between candidates on the basis of sex, origin or other characteristics?What do we, as an SME, have to arrange now if we use AI in HR?

Next step

Describe one AI tool, for example your recruitment or planning software, in four questions. You see at once which rules may apply, with the source, before you enter any data.

Map your AI tool

Looking for a specialist? We will look with you for a suitable specialist; mediation is not guaranteed.

This page gives general information based on the text of the law. It is not legal advice for your situation.