A Groundbreaking Law
The EU plans to create a legal framework for the development and use of artificial intelligence with its AI regulation. The Parliament is set to vote on the hard-fought draft today. The European AI regulation is the world's first attempt to introduce a comprehensive law to regulate artificial intelligence, making it a groundbreaking law. The EU Commission hopes that it will find imitators abroad, similar to the General Data Protection Regulation (GDPR).
However, formulating the regulation in a balanced manner was a difficult balancing act for legislators in Brussels. The parliamentary committees involved in the process struggled for 18 months to reach a position on the EU Commission's proposal.
Classification by Risk
According to the regulation, artificial intelligence should be classified according to the risks associated with its intended use, ranging from low-risk, moderately risky, to high-risk and prohibited AI.
Artificial intelligence capable of oppressing people should be entirely banned. This includes "social scoring" systems that evaluate people's behavior, automated emotion detection, such as during the interrogation of suspects, and widespread monitoring with biometric real-time data in public places.
However, according to the current draft, data can be accessed retroactively based on court order, if it is related to serious crimes. Dispute over Biometric Data
Even just before the vote, there is still disagreement in the parliament over the extent to which biometric data should be used in criminal investigations. Within the Christian Democratic European People's Party (EPP) faction, some members are insisting on changes to the draft.
The issue specifically concerns exceptions when using AI to search for missing children or to combat terrorism. Some EPP members hope that the countries will increase pressure on the parliament to make further changes in this regard. However, the Greens, the Left, and the Pirates are warning against mass surveillance.
ChatGPT would be Permitted
In addition to prohibited AI, there will be other categories of artificial intelligence. High, moderately risky, and low-risk AI, such as AI-powered toys, should generally be permitted. This also applies to so-called generative AI, such as the Chatbot ChatGPT, which can independently write articles using information collected from the Internet.
However, the principle is that the riskier the AI, the higher the requirements. Manufacturers must assess the risks of their products, meet certain training data standards, and be checked by inspection authorities. For example, the Social Democrats criticize the fact that companies conduct their risk analysis by themselves, and only after that are inspected by test authorities to see if their analysis is reasonable.
The regulation also aims to ensure that artificial intelligence does not rely on distorted datasets and potentially discriminate against individuals. For example, when AI checks creditworthiness or is used to acquire employees. Negatives examples of this practice can already be seen in the US, where black people are less likely to receive a loan.
Expectations and Concerns are High
Expectations are high for the European AI law, as are concerns. Sam Altman, CEO of OpenAI, which developed ChatGPT, warned of existential dangers posed by unregulated AI. Smaller developers, on the other hand, fear they won't be able to comply with the many documentation requirements. Civil rights activists hope that the AI law will not continue to promote discrimination and disadvantage.
If the Parliament passes its position at noon, negotiations with EU member states can begin. The rules for the use of artificial intelligence in law enforcement are likely to be a significant point of contention.
If an agreement can be reached before the next European elections in the coming year, the AI regulation is expected to take effect in 2026. However, artificial intelligence will continue to develop until then. As a result, it is likely that further AI regulations will follow in the European Union.