After the emergence of the chatbot ChatGPT, Brussels spent most of last year approving draft texts. EU member states on Friday approved landmark rules on controlling artificial intelligence, following difficult final-mile negotiations on what has been called the world's first legislation of its kind. Brussels first proposed AI legislation in 2021, but the process was greatly accelerated after the explosion of the chatbot ChatGPT. As early as December, EU countries and legislators had reached agreement on a set of draft rules, and their approval was expected to be a formality. However, some countries, including France and Germany, expressed concerns that the AI law would fail in the final stages, triggering a new round of negotiations to determine its content. After the concerns of France and Germany were alleviated, EU ambassadors finally signed the text at a meeting in Brussels on Friday. Belgium, the rotating presidency of the EU, said, "We are pleased to announce... that the ambassadors of the member states have just unanimously confirmed the final text." Thierry Breton, the EU's top technology enforcer, praised these "historic, world-first, groundbreaking" rules. He said, "The AI Act has unleashed a lot of passion... and it's right! Today, member states have supported the political agreement reached in December, recognizing the perfect balance negotiators found between innovation and security." Protecting Innovation EU law will regulate artificial intelligence systems based on risk assessments of relevant software models, which will be conducted by companies themselves. For example, the higher the identified risks to individual rights or health, the greater the obligations of the system. Berlin and Paris are particularly keen to protect AI startups to prevent these rules from hindering future "European AI champions." To address their concerns, the EU executive launched a series of measures at the end of January to support innovation and AI development by European startups. ChatGPT has demonstrated astonishing progress in a technology where generative AI can quickly generate text, images, and audio through simple commands in everyday language. The chatbot from OpenAI in the United States can compose rigorous articles and passionate poetry in seconds and pass medical and legal exams. The world is facing the dangers brought by artificial intelligence at a high speed, from the spread of false information online to manipulated videos and images. But it is also clear that generative AI, including Google's chatbot Bard, has unleashed huge new opportunities that Europe wants to take advantage of. Germany welcomes the agreement. Economy Minister Robert Habeck said Berlin will focus on "the convenience of innovation, the legal clarity for companies, and the necessary lightweight and non-bureaucratic structures" when implementing the law. German Digital Minister Volker Wissing congratulated himself earlier this week on the "improvements" to the text for "small and medium-sized enterprises," which will "avoid disproportionate requirements and ensure that we maintain international competitiveness." France also expressed concerns about copyright and generative AI, but an EU diplomat said the text struck the right balance between "protecting copyright and trade secrets."