MEPs agree on Negotiating Mandate for the Artificial Intelligence Act
After many months of negotiations, members of the European Parliament (MEPs) have found a compromise on the flagship legislative proposal on Artificial Intelligence (AI), the AI Act. The Act could become a global standard, similar to the GDPR which had major implications for firms worldwide.
The European law on AI was proposed by the Commission in April 2021 and could become the first AI law by a major regulator worldwide. After the Commission published its proposal, EU ministers green-lighted a general approach to the AI Act at the Telecom Council meeting on 6 December. On the side of the European Parliament, a joint Committee procedure was initiated. Consequently, two parliamentary committees, namely the Committee on the Internal Market and Consumer Protection (IMCO) and the Committee on Civil Liberties, Justice and Home Affairs (LIBE) are equally responsible for the work on the parliamentary side with five further Committees selected to provide their opinions. This choice is based on the fact that the topic of AI has links with numerous political domains. The complexity of the matter addressed, led to a lengthy decision-making process in the parliament. The process was also affected by discussions caused by the recent launch and gain of popularity of ChatGPT.
The proposal’s goal is to make AI safe and trustworthy. Most importantly, the law assigns applications of AI to three risk categories. Firstly, applications and systems that create an unacceptable risk, such as government-run social scorings are banned. Secondly, high-risk applications are subject to specific legal requirements. Lastly, applications not explicitly banned or listed as high-risk are largely left unregulated. The Prohibited practices and AI solutions labelled as high-risk were subject to intense debates in the European Parliament over the past months.
The negotiators in the European Parliament suggest several changes to the proposed text. One proposal to prohibit AI-powered tools for general monitoring of interpersonal communications was dropped due to opposition from the conservative European People’s Party (EPP). In exchange, the ban on biometric identification software was extended to cover ex-post use only for serious crimes with pre-judicial approval. The EPP, which has a notable law enforcement faction, accepted not to table alternative amendments that might threaten support for the entire file but might still try to change the ex-post biometric ban. The regulation also prohibits purposeful manipulation and emotion recognition AI-powered software is banned in law enforcement, border management, workplace, and education. The ban on predictive policing was extended from criminal offences to administrative ones based on a child benefit scandal in the Netherlands. Providers of high-risk AI solutions must comply with stricter requirements, and certain AI models that pose a significant risk of harm to health, safety, or fundamental rights are classified as high risk. Extra safeguards were included for the processing of sensitive data to detect negative biases, and general principles were proposed to apply to all AI models. High-risk AI systems must, furthermore, comply with European environmental standards and record their environmental footprint.
The report could still be subject to small adjustments made at the technical level ahead of the vote in the joint committee scheduled on 11 May. The report is expected to go to plenary in mid-June. If approved, interinstitutional negotiation between Parliament, Council and Commission, will start after.