European Parliament Proposes Bill to Regulate AI in EU
The European Parliament has put forward a bill for discussion that may become the standard for regulating artificial intelligence (AI) in the European Union (EU). After the law comes into force, organizations that do not comply with it may face fines of € 40 million or 7% of their total turnover.
In addition to fines, the bill proposes bans on systems of remote biometric identification in real-time in public places and systems that categorize people into groups based on their race and ethnicity. The latter would prevent the implementation of systems such as predictive protection and recognition of emotions.
The bill also makes developers and users responsible for compliance with the requirements of the law based on risk assessment. The document contains a definition of “high risk” in terms of health and safety or basic rights. Developers of high-risk systems will have to meet mandatory requirements for reliable AI and follow procedures for assessing compliance before the systems can be placed on the EU market.
The commission aims to create a system for registering AI autonomous enterprises with high risk in a publicly available EU database. The law imposes an obligation on AI developers to evaluate the quantity and suitability of training data sets and check them for possible prejudices.
The European Parliament will discuss the bill today, and a vote will be held on Wednesday. However, it is important to note that the European Parliament passed the world’s first rules for AI on May 11 (AI ACT) aimed at ensuring AI’s security, transparency, and ethics in Europe. The rules establish obligations for suppliers and users of AI, depending on the risk level that AI may create.
While the new bill addresses concerns about regulating AI in Europe, some experts worry that it could undermine the potential for innovation and the competitiveness of the European AI sector. It could also create barriers to cooperation and knowledge exchange with American partners and contribute to the development of unsafe AI. Therefore, the bill requires more discussion and clarification before its final adoption.
In October 2020, the White House published a Bill on the Rights of Artificial Intelligence that is expected to regulate the development, use, and deployment of automated systems in the AI era. The bill aims to regulate AI based on the fifth basic principle.