Prohibited AI Practices in Healthcare under the European Artificial Intelligence Act.

Area of Science:

• Healthcare AI Regulation
• AI Ethics
• Public Health Policy

Background:

• The EU's Artificial Intelligence Act establishes an "unacceptable risk" category for AI.
• Article 5 prohibitions within the Act have received less attention than high-risk AI regulations.
• Healthcare and public health are rapidly adopting AI, presenting unique ethical challenges and vulnerabilities.

Purpose of the Study:

• To analyze the implications of the EU AI Act's prohibitions on healthcare and public health applications.
• To examine how specific AI technologies in health may be banned under Article 5.
• To assess the impact of medical and safety exceptions on AI vulnerability protections in healthcare.

Main Methods:

• Legal analysis of the EU AI Act and Article 5 prohibitions.
• Interpretation of the European Commission's 2025 interpretative Guidelines.
• Integration of legal analysis with real-world health examples and AI applications.

Main Results:

• Several AI health applications, including emotion recognition and biometric categorization, may be prohibited.
• AI targeting vulnerable populations in healthcare settings could fall under the bans.
• Medical and safety exceptions in the Act risk undermining protections for vulnerable individuals.

Conclusions:

• The EU AI Act's prohibitions have significant implications for AI in healthcare and public health.
• Existing exceptions may weaken essential safeguards for vulnerable populations.
• A framework is proposed for interpreting AI prohibitions to guide ethical boundaries in health AI.