Screen an AI system or product feature against the AI Act Article 5 prohibited practices, including the prohibitions added by the AI Omnibus

Documented steps

1. Confirm scope and timing: Article 5 prohibitions apply irrespective of high-risk status and have applied since 2 February 2025. They were not postponed by the AI Omnibus. Act text: https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32024R1689
2. Pull the baseline list from Article 5(1) of Regulation (EU) 2024/1689 and screen the system against each category: subliminal, purposefully manipulative or deceptive techniques materially distorting behaviour and causing significant harm; exploitation of vulnerabilities arising from age, disability or a specific social or economic situation; social scoring leading to detrimental or disproportionate treatment; predicting the risk of a natural person committing a criminal offence based solely on profiling or personality traits; untargeted scraping of facial images from the internet or CCTV to build or expand facial recognition databases; inferring emotions in the workplace or in education institutions outside the medical and safety exceptions; biometric categorisation to infer race, political opinions, trade union membership, religious or philosophical beliefs, sex life or sexual orientation; and real-time remote biometric identification in publicly accessible spaces for law enforcement outside the narrow statutory exceptions.
3. Screen additionally against the prohibitions inserted by the AI Omnibus, which cover AI systems generating or manipulating material depicting an identifiable person's intimate parts or sexually explicit conduct without that person's consent, and AI systems generating or manipulating child sexual abuse material. Read the inserted points verbatim in the OJ text: https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=OJ%3AL_202601744
4. Confirm the application date of the newly inserted prohibitions in the final application article of the amending regulation at the same URL. The Commission's overview FAQ lists these among the banned uses but does not restate a separate date: https://digital-strategy.ec.europa.eu/en/faqs/navigating-ai-act
5. Do not rely on the Commission's per-article explainer pages alone for Article 5 — as of early August 2026 several carried explicit disclaimers that they had not yet been updated for the AI Omnibus amendments.
6. On a positive hit, stop placing on the market, putting into service and using the feature. Article 5 practices cannot be mitigated into compliance the way high-risk systems can; the remedy is removal or redesign that eliminates the prohibited element, for example genuine, specific, informed and explicit consent capture where consent is the operative distinction.
7. Escalate borderline cases to legal and to the data protection officer, since biometric categorisation and emotion inference overlap heavily with GDPR special-category processing.
8. Document the screen — the analysis, the exemption relied on, and the decision — as the evidence file for any market surveillance investigation.
9. Size the exposure: Article 99(3) sets administrative fines for Article 5 breaches at up to EUR 35 000 000 or, for an undertaking, up to 7% of total worldwide annual turnover for the preceding financial year, whichever is higher. Text: https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-99
10. For SMEs including start-ups, apply Article 99(6): the cap is whichever of the percentage or the fixed amount is LOWER, reversing the usual rule.
11. Re-run the screen on every material feature change, new modality (especially image, video and audio generation), and new deployment context.