The EU AI Act went into enforcement on August 2, 2026, imposing immediate bans on certain high‑risk AI uses while delaying other compliance obligations to December 2027.

The European Union’s AI Act entered its enforcement phase on 2 August 2026, triggering an immediate ban on a narrow set of AI applications deemed “unacceptable risk” while postponing broader high‑risk compliance requirements until December 2027.

Immediate Prohibitions Under the Act

The law explicitly bars the use of AI systems that manipulate human behaviour, exploit vulnerabilities of specific groups, or enable real‑time remote biometric identification in public spaces without a clear legal basis. Operators of such systems must cease processing within a short grace period or face heavy penalties.

High‑Risk AI Rules Deferred

Unlike the swift bans, the Act’s extensive high‑risk regime—covering AI for critical infrastructure, recruitment, credit scoring, and medical diagnostics—has been given a transition window that runs until the end of 2027. During this time, providers must conduct conformity assessments, maintain detailed documentation, and implement post‑market monitoring, but they are not yet subject to the full enforcement regime.

The staggered approach aims to give businesses time to adapt to the rigorous standards while protecting citizens from the most dangerous applications from day one.

Enforcement Mechanisms and Penalties

National supervisory authorities will issue formal notices to violators of the prohibition rules, and non‑compliance can lead to fines of up to 6 % of global annual turnover or €30 million, whichever is higher. For delayed high‑risk obligations, authorities will monitor progress through periodic reporting and on‑site inspections after the 2027 deadline.

  • Immediate bans target AI‑enabled deep‑fakes, covert biometric surveillance, and manipulative recommendation engines
  • High‑risk AI categories such as medical diagnostics, credit scoring, and recruitment are subject to a 2027 compliance deadline
  • Fines can reach up to 6 % of global turnover or €30 million for serious breaches
The phased rollout reflects a pragmatic balance between rapid consumer protection and giving the industry a realistic timeline to meet the Act’s technical standards.

Stakeholders have welcomed the clear separation, noting that the early bans address the most pressing societal harms while the extended timeline for high‑risk AI provides a workable path for compliance across the EU’s diverse market.

Winzheng analysis of EU AI Act enforcement timing