The plenary session of the European Parliament is scheduled to vote on Wednesday to start negotiations to approve the first global regulation on artificial intelligence (AI) systems, with the aim of curbing potential damage without blocking innovation. The European Union’s (EU) initial plans to regulate the use of AI date back to 2021, but the issue became urgent after the emergence of ChatGPT, which highlighted the rapid development of the sector.
Once it gains approval from the European Parliament, EU officials hope to immediately begin direct negotiations with all 27 member countries, with the idea of having an agreement finalized by the end of the year. However, even if this goal is achieved, the regulation would not come into force until 2026, so the EU will also need to negotiate an interim, voluntary framework with technology companies developing AI.
A majority of MEPs support this draft regulation as an important milestone. It is considered the right time for Europe to regulate AI because of its profound impact, as MEP Dragos Tudorache highlighted during the parliamentary debate in Strasbourg. Margrethe Vestager, vice president of the European Commission, noted that the aim is to create trust and legal certainty to enable the positive development of AI.
The proposed law will regulate AI according to the level of risk, establishing greater obligations for systems that pose a higher risk to people’s rights or health. The EU’s proposed high-risk list includes AI used in critical infrastructure, education, human resources, public order, and migration management.
The European Parliament has added additional conditions before classifying something as high-risk, including the potential to harm health, safety, fundamental rights, or the environment. Special requirements are also set for generative AI systems, such as ChatGPT and DALL-E, which can generate text, images, code, audio, and other media, including the obligation to inform users that the content was produced by a machine and not a human being.
European lawmakers and scientists have highlighted the need for a common approach to address the risks of AI and have proposed banning AI systems for biometric surveillance, emotion recognition, and predictive surveillance. In addition, they have called for a moratorium on the development of the most powerful systems until they are better regulated by law, in view of the risks of manipulating public opinion and the dangers to democracy, as seen with the spread of fake images created by AI-based applications.