What do stakeholders think about the EU AI Act? EU Commission publishes feedback

The European Commission has published an analysis of stakeholder feedback to the EU AI Office on parts of the EU AI Act - specifically the definition of AI and prohibited practices which have been in application since 2 February 2025.

The aim of the feedback is to help improve the EU's guidance documents on the definition of AI (here) and prohibited practices (here).  Both are expected to evolve over time in response to practical experience, emerging use cases, and new questions.  

Key points of feedback include:

• Definition of AI:
  • there remains significant concern about the ‘clarity and scope of the definition of “AI system” in the EU AI Act, with stakeholders strongly emphasising the need for more precise definitions of terms like "adaptiveness," "inference," and "autonomy." A major concern was the potential for current definitions to inadvertently include traditional software systems that do not exhibit true AI characteristics.’
• Prohibited practices:
  • clearer guidelines are needed to help distinguish prohibited applications and legitimate uses cases of emotion recognition, biometric categorisation, and social scoring;
  • clearer thresholds for what is meant by ‘manipulation’ and ‘significant harm’ in AI systems;
  • clearer guidance is needed for how the AI Act and GDPR interact.
• there continues to be concern about compliance challenges for small and medium enterprises

The report recognises that the demographics of respondents may mean the results do not fully represent those most affected by AI systems, although it appears that responses were provided by a range of types of respondent (e.g. civil society, academia, citizen, industry), from across sectors, and various EU and non-EU jurisdictions.

The report e.g.was produced in conjunction with the Centre for European Policy Studies and can be found here, and includes a detailed breakdown of respondents and their responses (subject to redactions).

If you would like to discuss how current or future regulations impact what you do with AI, please contact Tom Whittaker, Brian Wong, Lucy Pegler, Martin Cook, Liz Smith or any other member in our Technology team.  For the latest on AI law and regulation, see our blog and newsletter.