On 14 May 2023, the European Parliament adopted its position on the Act on Harmonised Rules for Artificial Intelligence (AI Act), including quality of service requirements for "high-risk" AI systems. The AI Act is based on a risk-based management approach and would establish obligations for providers and users depending on the level of risk generated by the AI system, namely an "unacceptable" risk, "high-risk", and "low or minimal" risk. In particular, under the Parliament position, the definition of "high-risk" AI systems is expanded to include systems that could pose or cause harm to the health and safety of individuals or could have a negative impact on individuals' fundamental rights and influence voters in political campaigns. The high-risk systems require a management system and testing to ensure it performs consistently with their intended purpose. Such systems' providers must develop security systems to ensure that natural persons can effectively oversee them. The Parliament will start the negotiation with the Council on the final version of the Act.
Original source