European Commission (EC) — European AI Office · European Union
EU AI Act Annex III(1) — Biometric Identification and Categorization
EU-directed marketing must not offer prohibited biometric AI uses (sensitive-trait categorisation, workplace or school emotion recognition, untargeted face scraping) or misstate the high-risk status of biometric systems.
Regulation (EU) 2024/1689 Arts 5(1)(e)-(g), 6, 48, Annex III(1), as amended by 2026/1744
What is EU AI Act Annex III(1) — Biometric Identification and Categorization?
EU-directed marketing must not offer prohibited biometric AI uses (sensitive-trait categorisation, workplace or school emotion recognition, untargeted face scraping) or misstate the high-risk status of biometric systems. Apply to EU-directed marketing for biometric identification, categorisation and emotion-recognition AI products. This guide summarizes the supplied rule information for marketing review. Use the linked official source to confirm the current text, applicability
general_audience, general_adults_18plus, customers, business, small_business · All Industries · ai genai
Where does this rule apply?
European Union
Which marketing channels are covered?
all_channels
When does this rule take effect?
2026-05-12
What does this rule require?
EU-directed marketing must not offer prohibited biometric AI uses (sensitive-trait categorisation, workplace or school emotion recognition, untargeted face scraping) or misstate the high-risk status of biometric systems.
How can marketing teams comply?
Apply to EU-directed marketing for biometric identification, categorisation and emotion-recognition AI products.
What is EU AI Act Annex III(1) — Biometric Identification and Categorization?
EU-directed marketing must not offer prohibited biometric AI uses (sensitive-trait categorisation, workplace or school emotion recognition, untargeted face scraping) or misstate the high-risk status of biometric systems.
Who does this rule apply to?
general_audience, general_adults_18plus, customers, business, small_business · All Industries · ai genai
Article 5 bans placing on the EU market or using certain AI practices, including manipulative or vulnerability-exploiting techniques, social scoring, workplace and school emotion recognition, sensitive biometric categorisation, untargeted facial scraping and solely profiling-based crime prediction; the Digital Omnibus adds a ban on AI generating CSAM and non-consensual intimate imagery from 2 Dec 2026.
AI systems used for Annex III purposes such as credit scoring, life and health insurance pricing, hiring, education and biometrics are high-risk; under the Digital Omnibus their obligations apply from 2 Dec 2027 (Annex I products from 2 Aug 2028), and marketing must not misstate their regulatory status.
Under the EU UCPD, traders must not present AI-generated or fictional people, reviews or product depictions as genuine consumer experiences or real results, and must explain how they verify reviews claimed to come from real buyers.
Credit ad imagery must not signal that applicants on a prohibited basis are unwelcome, and translated ads must carry the same material terms and disclosures as the English version; diverse imagery is not legally required.
Job ads may not indicate a preference, limitation or discrimination based on race, color, religion, sex or national origin, except a stated bona fide occupational qualification or a religious organization's religious preference.
EU job advertising must not indicate preferences based on age, sex, race or ethnic origin, religion or belief, disability or sexual orientation, and, where the Pay Transparency Directive is transposed, job ads must be gender-neutral and must not ask about pay history.
Request a free compliance scan of your homepage and book a time with Warrant.