EU unveils recommendations to rein in powerful AI models

On Thursday, the European Union introduced a set of long-anticipated guidelines aimed at regulating cutting-edge AI models, including OpenAI's ChatGPT. This move comes as companies brace for compliance with the EU's comprehensive new AI legislation, which is set to take effect for general-purpose AI systems on August 2. The EU has faced significant pressure to postpone the enforcement of this landmark law, which gradually rolls out various obligations. As it works to enhance its competitive edge against the United States and China, many European tech firms, as well as some major U.S. tech companies, are advocating for a slowdown in implementation. Despite these calls for delay, the European Commission, acting as the bloc's digital overseer, has firmly resisted the idea. The commission has now released a code of practice for these AI systems, crafted with contributions from independent experts and input from the AI model providers. This code outlines several practical recommendations, such as prohibiting the use of data from known piracy websites. The guidelines pertain to general-purpose AI models, which include Google's Gemini, Meta's Llama, and Elon Musk's Grok chatbot, the latter of which has recently faced backlash for making antisemitic remarks. Under the new law, developers of these models are required to disclose the types of content, such as text or images, used in training their systems, ensuring compliance with EU copyright regulations. Originally scheduled for release in May, the code faced accusations of being diluted under industry influence. Organizations like Corporate Europe Observatory and Lobby Control claimed that Big Tech significantly impacted the process to weaken its effectiveness. Before companies can voluntarily adopt the code, it must receive approval from EU member states. The European Commission has stated that businesses signing onto the code will experience a lighter administrative load and greater legal clarity compared to those demonstrating compliance through other means. Recently, nearly 50 major European corporations, including Airbus, ASML, Lufthansa, and Mercedes-Benz, called for a two-year moratorium on the new rules, warning that the EU's intricate regulations could jeopardize the bloc's AI ambitions and hinder the emergence of European tech leaders. The EU is set to enforce the new regulations for general-purpose AI models one year after August 2 for new models, while existing models will have until August 2027 to comply.