The future of the Republic’s groundbreaking Artificial Intelligence Act hangs in the balance as the three largest economies in Europe oppose regulating advanced forms of artificial intelligence.
France, Germany, and Italy are obstructing negotiations on a contentious aspect of the EU’s AI legislation to prevent Europe from impeding the development of foundational AI networks that support large-language models like OpenAI’s GPT and Google Bard.
Government officials argue that relaxing regulations on these cutting-edge models could jeopardize the EU’s position as a leader in AI technology.
In a published document shared with other EU governments and obtained by POLITICO, the three influential German entities advocate for a regulatory framework that promotes innovation and competition to empower European players in the global AI race. They propose self-regulation of foundational models through corporate commitments and codes of conduct.
This stance pits them against German politicians who are adamant about restricting foundational designs, creating tension within the Franco-German-Italian alliance.
A confidential source involved in European Parliament negotiations referred to the situation as “a declaration of war,” highlighting the intensity of the disagreement.
The deadlock could potentially halt discussions on the Artificial Intelligence Act altogether. Following the resistance from the three countries to regulate foundational designs, negotiations at the European level have reached an impasse.
With the deadline of December 6 looming and the European Parliament facing reelection in June 2024, diplomats are under pressure to find a resolution before time runs out.
The push to relax the regulatory framework in Europe contradicts the prevailing belief that the technology sector requires stricter oversight, catching many by surprise. This shift comes at a time when industry leaders are advocating for robust regulations, and even the United States is outlining its regulatory agenda through an extensive Executive Order on AI.
Renowned AI expert Yoshua Bengio warned against ignoring foundational models, emphasizing the importance of regulating these advanced systems to avoid potential risks and ensuring a level playing field in the AI landscape.
The European Union generally supports stringent regulations on AI applications, especially in sensitive areas like healthcare, immigration, and employment. However, the diverse capabilities of foundational models pose challenges in assessing their potential risks.
German lawmakers proposed stringent requirements for developers of foundational models, including third-party model testing, regardless of the model’s intended use. This proposal contrasts with the two-tier approach rejected by the three governments, which would have imposed additional obligations based on the processing power of the models.
Despite some EU nations advocating for a broader scope in the AI Act, the Council is currently aligned with the position of the three influential countries, making it challenging to deviate from their stance.
