The European Union has an ongoing debate over the first laws to regulate AI. Leaders like … [+]
European Parliament negotiators reach agreement consensus last Friday on the wording of the long-debated AI law, the first major legislation on artificial intelligence in the European Union. The bill emerges amid growing concerns among EU member states about the impact the regulations will have on innovation. As a result, the tentative agreement could still fail due to infighting.
The AI Act’s journey began about two and a half years ago, when the first version of the law was made public in 2021. This was before the release of cutting-edge AI models like ChatGPT. With the rise of “core models” – general-purpose AI systems capable of a wide range of applications – has come a shift in the EU’s legislative approach towards a focus on the most advanced technologies . These “general purpose” AI models will have their own sets of tailored rules.
The focus on foundation models has been one of the areas that has sparked the most intense debate. Emmanuel Macron of France criticized the bill, argue this could kill innovation on the continent. The language still needs to be approved by EU member countries and the European Parliament before it becomes law, a process that could take months. In the face of attacks from heads of state like Macron, the AI law could still easily be derailed, creating considerable uncertainty.
As the law currently stands, AI applications will be categorized by risk, with AI systems “high risk.” submitted to the strictest rules. The requirements include pre-market testing, mandatory risk mitigation systems, standards surrounding the datasets used to train AI systems, and human oversight.
For now, businesses will have to self evaluation whether their designs meet the compliance threshold for general purpose foundation designs. Models will initially be ranked based on the computing power required to run them, but this is subject to change as more metrics and benchmarks become available.
The AI law also introduces a number of transparency measures, including the obligation for companies to inform users when they interact with AI systems such as chatbots or emotion recognition systems, the requirement to label deepfakes and other AI-generated content, as well as various other disclosure and documentation requirements. . European copyright law must also be respected.
The law exempts the most mundane AI applications from strict rules, leaving a significant portion of AI technologies relatively free from regulatory oversight. However, he prohibitions Uses of AI are deemed to pose “unacceptable” risks. These include biometric systems that identify individuals based on sensitive biological data, retrieval of facial images from online databases, emotion recognition systems at work or in schools, social rating systems as they exist in ChinaAI systems that manipulate human behavior and most forms of predictive policing.
A source of contention has come from countries that see the potential of using AI in the fight against terrorism. To address these concerns, the bill gives countries considerable latitude to use AI in ways national security context. However, questions remain related to the operational and compliance burdens imposed on businesses under the law. The concerns echo those surrounding Europe’s data privacy law, the General Data Protection Regulation, which some say has led to a competitiveness gap between the American and European technology sectors.
The bill also creates a new European AI Office for enforcement and a strict sanctions system with fines ranging from 1.5% to 7% of a company’s global turnover. In the future, the AI law is unlikely to come into force until 2026, offering companies a two-year deadline to comply. On the one hand, this creates flexibility for businesses. On the other hand, the AI landscape has changed so much over the past two years that the new law could easily become obsolete before it takes effect.
Supporters of the AI Act hope it will set the global standard for AI regulation. They should be careful what they wish for. If the new law stifles innovation in the way many expectthis further relegates the EU to the technological gap, just like all countries that decide to follow his advance.