Citing the increased interest in AI applications, the European Union’s so-called
“AI Act” is intended to clarify what AI software may and may not do, what obligations companies have and what requirements products must meet.
The AI Act (AIA for short) is part of the AI package presented by the EU Commission in April 2021. Last December, the EU Parliament and the EU Council reached a political agreement on the proposed act.
The Artificial Intelligence Act provides for a risk-based approach. AI systems with unacceptable risks are prohibited, while high-risk systems are assigned to specific categories. Under the AI Act, AI systems with limited risk only have to meet minimal transparency requirements.
Medical device software is assigned to the category of high-risk systems. One consequence of this is that a notified body must in any case be involved in the conformity assessment procedure.
On December 9, 2023, Parliament reached a provisional agreement with the Council on the AI Act, which now has to be formally adopted by both parties to become EU law. The final steps are to take place in the coming meetings of Parliament’s Internal Market Committee and Civil Liberties Committee.
The draft act can be viewed here: https://eur-lex.europa.eu/legal-content/DE/TXT/HTML/?uri=CELEX:52021PC0206