The European Union reached an agreement on a law to regulate artificial intelligence, setting it apart from other countries worldwide. The Artificial Intelligence, or AI Act, will establish regulations for the safe and effective development and use of these technologies.
The act will make the EU the first to establish a comprehensive set of rules addressing a wide range of AI applications, including facial recognition, social scoring, and Artificial Intelligence that can influence human conduct.
In addition to a risk-based approach, companies that use AI will be required to provide transparency regarding how their systems operate and the data they utilize. It will also be necessary to have human oversight in place for their AI systems that present a significant risk to the safety and well-being of individuals. Companies will not be able to utilize AI in a discriminatory way against specific individuals or groups and will be held accountable for any harm caused by their systems.
This legislation represents a significant step forward in the regulation of AI, which is expected to profoundly impact the world economy and society in the years to come. The EU has taken the lead in addressing the potential societal risks of AI while fostering the continued development of these technologies.
One of the most controversial aspects of the legislation is whether it should apply to general-purpose AI and so-called foundation models. These models are trained on massive datasets and underpin a broader range of AI applications. The legislation will require that these systems comply with a set of transparency regulations, including compliance with EU copyright law and the creation of detailed summaries of the content used to train AI models.
The deal also bans specific AI applications, such as untargeted scraping of images to create facial-recognition databases. It sets rules for systems that are considered to be high-risk, and penalties for breaking the rules could reach up to 7% of a company’s global revenue.
“Europe has positioned itself as a pioneer, understanding the importance of its role as global standard setter,” Thierry Breton, the European commissioner who helped negotiate the deal, said in a statement.
However, some critics have argued that the rules are too strict and could stifle innovation. “The AI race is not one Europe can miss out on,” said Cecilia Bonefeld-Dahl, the director-general of DigitalEurope.
The first draft of the AI Act was released in 2021. But policymakers found themselves rewriting the law as technological breakthroughs emerged. The initial version did not mention general-purpose AI models like those that power ChatGPT.
Overall, the EU’s AI Act is a significant step forward in regulating AI. It will help ensure that AI is developed and used responsibly and ethically. It is setting a global standard for other countries to follow.
The legislation still needs final approval from parliamentarians and representatives from the EU’s 27 member states and is unlikely to take full effect before 2026.