EU AI Act Full Enforcement Begins August 2 — Facial Recognition Bans, High-Risk AI Rules Binding

Correction (2026-08-06): the high-risk AI system obligations described below did NOT take binding effect as scheduled. The Digital Omnibus on AI — adopted by the European Parliament on June 16, 2026 and the Council of the EU on June 29, 2026, and reported as entering into force in early-to-mid July 2026 — pushed the high-risk deadline (employment, credit, law enforcement, critical infrastructure AI) to December 2, 2027 for stand-alone systems (August 2, 2028 for AI embedded in regulated products) — see 2026-07-27–eu-ai-omnibus-delays-high-risk-ai-deadline-to-december-2027 for the corrected timeline. The facial-recognition bans and transparency requirements below DID take effect as scheduled August 2; only the high-risk-system compliance deadline moved.

The EU AI Act becomes fully enforceable on August 2, 2026, activating binding requirements for high-risk AI systems, banning real-time public-space facial recognition (with narrow exceptions), and classifying police post-incident facial recognition as high-risk — requiring court approval and documentation rather than banning the technology outright — while prohibiting untargeted facial image scraping, the Clearview model specifically. Employment AI restrictions also take effect. No equivalent US federal framework exists; the transatlantic divergence is already shaping procurement decisions.

The EU framework’s enforcement lands in the same week Clearview AI secured FedRAMP “In Process” status at the High impact level in the US (2026-07-08–clearview-fedramp-cbp-facial-recognition-all-travelers) — the same surveillance model the EU is banning is simultaneously being cleared for streamlined federal procurement in the US, marking a concrete fork in Western biometric-surveillance governance rather than a rhetorical one.

Corrected 2026-10-01: corrected “prohibiting police post-incident facial recognition” to “classifying police post-incident facial recognition as high-risk” — the cited State of Surveillance source states police “can record crowds and run facial recognition later” subject to court approval, “but the technology itself isn’t banned.” Also corrected the regulation’s name from “AI Act Omnibus” (regulation 2026/1744) to “Digital Omnibus on AI” and its force-date from July 27 to early-to-mid July 2026 — the cited Tech Times source names it the “Digital Omnibus on AI,” cites no regulation number, and reports it adopted by Parliament June 16 and the Council June 29, 2026, entering into force “this week” as of its July 10, 2026 publication.