Key Developments

The EU Council has agreed on significant amendments to streamline the AI Act implementation, introducing more proportionate rules and harmonised enforcement across member states. Most notably, the Council has added new prohibited AI practices, specifically banning the generation of non-consensual sexual and intimate content or child sexual abuse material using AI systems.

The updated timeline extends key compliance deadlines: standalone high-risk AI systems must comply by 2 December 2027, while high-risk AI systems embedded in products have until 2 August 2028. However, the most demanding obligations for high-risk systems still take effect on 2 August 2026.

A landmark development is the publication of the first draft Code of Practice on AI content transparency, proposing an “EU common icon” to help users instantly identify AI-generated or AI-edited content depicting real events or people.

Industry Context

These changes reflect the EU’s attempt to balance innovation with protection, responding to industry concerns about implementation complexity while strengthening safeguards against AI misuse. The extended deadlines provide breathing room for companies developing high-risk AI systems, particularly important for Irish and European AI firms navigating compliance.

The new prohibited practices directly address growing concerns about deepfakes and non-consensual intimate imagery, issues that have escalated with improved AI generation capabilities. The transparency requirements signal Europe’s commitment to maintaining public trust in AI systems.

Practical Implications

For Irish AI developers and European companies, the extended deadlines offer additional time to ensure compliance with high-risk system requirements. However, August 2026 remains the critical milestone for core obligations.

The EU common icon requirement will impact any AI system generating visual content, requiring implementation of detection and labelling mechanisms. Companies should begin preparing transparency measures now, with supporting guidelines expected in Q2 2026.

The European Commission’s recruitment of AI technology specialists suggests increased enforcement capability, indicating serious implementation intent.

Open Questions

Key uncertainties remain around the technical implementation of the EU common icon system and how detection mechanisms will work across different AI models. The practical enforcement of non-consensual content prohibitions across member states also requires clarification, particularly regarding cross-border cases and platform responsibilities.

Source: artificialintelligenceact.eu