Applies now · more from 2 Dec 2026Article 5 — banned outright
What it covers: A short, specific list of practices banned outright — social scoring, manipulative or subliminal techniques designed to cause harm, exploiting people’s vulnerabilities, most real-time remote biometric identification in public spaces for law enforcement, and untargeted scraping of facial images to build recognition databases.
What this looks like for an Irish SME: Realistically, almost nothing in ordinary business use. This tier targets specific, deliberate practices — not everyday AI-assisted admin, marketing or customer service.
What’s required: These practices simply can’t be used. Most of this list has applied since 2 February 2025; two further prohibitions take effect on 2 December 2026.
Maximum penalties for prohibited practices and high-risk failures can reach up to €35 million or 7% of global annual turnover. That figure gets a lot of headlines — but it applies to this narrow bracket of practices, not to a typical SME using AI for everyday admin.