News EU AI Act Implementation Milestones 2026

From AI Law Wiki
Jump to navigation Jump to search

The EU AI Act continues its phased implementation in 2026, with major enforcement provisions set to take effect on August 2, 2026. Key institutional and regulatory milestones have been reached in preparation for the August deadline, including the European Parliament's adoption of its negotiating position on Digital Omnibus amendments to the Act.

Prohibited Practices and AI Literacy (Effective February 2025)

Prohibited AI practices and AI literacy obligations have been in effect since February 2, 2025. Eight categories of prohibited practices include harmful manipulation, social scoring, and real-time biometric identification in public spaces for law enforcement purposes.[1] However, no formal enforcement actions identifying prohibited practices have been reported as of April 2026, since national market surveillance authorities' ex-post checks and fining powers do not activate until August 2026.[2]

European Parliament Position on Digital Omnibus Amendments (March 26, 2026)

On March 26, 2026, the European Parliament plenary adopted its negotiating position on proposed Digital Omnibus amendments to the EU AI Act, advancing to trilogue negotiations with the Council (which adopted its mandate on March 13, 2026) and the Commission.[3][4]

Key amendments proposed by Parliament include:[3]

  • Fixed application dates: Replaces the Commission's flexible backstop dates with fixed dates to eliminate regulatory uncertainty about when rules take effect.
  • Ban on non-consensual AI-generated intimate imagery: Introduces a prohibition under Article 5 on AI systems generating realistic sexually explicit images or videos of identifiable individuals without consent, aligning with a similar Council proposal.
  • Shortened transparency grace period: Reduces the compliance period for Article 50(2) marking obligations (for pre-August 2, 2026 systems) from six months to three months (to November 2, 2026).
  • Reinstated registration requirements: Retains EU database registration for self-assessed non-high-risk AI systems under Article 6(3), rejecting the Commission's proposed removal and aligning with the Council position.
  • Stricter data processing thresholds: Restores "strict necessity" standard for processing special categories of personal data in bias detection, limited to high-risk systems with exceptional extensions requiring necessity, proportionality, and links to health, safety, rights, or discrimination.
  • Annex I restructuring: Deletes Section A and moves New Legislative Framework legislation to Section B, altering the AI Act's interplay with sectoral product rules.
  • AI Office enhancements: Strengthens supervision over general-purpose AI models with resourcing requirements, while preserving national authority competence through exceptions.
  • Ban on "nudifier apps": Explicitly prohibits AI tools designed to generate non-consensual intimate imagery.

Parliament and Council show broad alignment on rolling back Commission simplifications, supporting fast trilogue negotiations ahead of the August 2, 2026 application deadline.[3]

EDPS Compass for AI Act Supervision

On March 17, 2026, the European Data Protection Supervisor (EDPS) issued "Towards Trustworthy AI in the EU Public Administration: The EDPS Compass for its New Role Under the AI Act, 2026–2027," outlining supervision and enforcement strategies for EU institutions under the Act.[5]

European Commission Enforcement Responsibilities

On March 31, 2026, an overview of enforcement provisions relating to Chapter V of the EU AI Act was published, detailing Commission responsibilities including consultations on national measures (Article 81) and fines for general-purpose AI (GPAI) model providers (Article 101). These powers activate from August 2, 2026.[6]

Member State Preparedness

Member States were required to designate national authorities by August 2, 2025, but only eight single points of contact had been listed by March 2026.[2] Each Member State must establish at least one AI regulatory sandbox by August 2, 2026, with no completions reported as of April 2026.[7]

August 2026 Deadline

The following provisions take effect on August 2, 2026:

  • High-risk AI systems listed in Annex III must comply with full requirements
  • Transparency obligations under Article 50 take effect
  • National market surveillance authorities gain enforcement and fining powers
  • Fines of up to EUR 35 million or 7% of global annual turnover become applicable[8][9]

GPAI Model Obligations

Full GPAI compliance obligations take effect on August 2, 2026, except for GPAI models placed on the market before August 2, 2025, which have until August 2, 2027. The Commission's AI Office has been issuing soft instruments including codes of practice. A proposed "digital omnibus on AI" (November 2025) seeks further centralization for GPAI-based systems.[2][10]

References

  1. ↑ European Commission, "Regulatory Framework for AI"
  2. ↑ 2.0 2.1 2.2 European Parliamentary Research Service, "Enforcement of the AI Act" (March 18, 2026)
  3. ↑ 3.0 3.1 3.2 Global Policy Watch, "MEPs Adopt Joint Position on Proposed Digital Omnibus on AI"
  4. ↑ European Parliament, "TA-10-2026-0098_EN"
  5. ↑ Artificial Intelligence Act Update Tracker
  6. ↑ ArtificialIntelligenceAct.eu, "Responsibilities of the European Commission / AI Office"
  7. ↑ ArtificialIntelligenceAct.eu, Implementation Timeline
  8. ↑ Barr Advisory, "EU AI Act 2026"
  9. ↑ EU AI Act Service Desk, "Timeline of Implementation of the EU AI Act"
  10. ↑ EPRS, "Enforcement of the AI Act" (ATA 785670, March 2026)
Retrieved from "https://ailawwiki.com/index.php?title=News_EU_AI_Act_Implementation_Milestones_2026&oldid=256"