France Raises Concerns Over the EU’s AI Act Despite Mounting Criticism
France continues to voice its reservations towards the European Union’s (EU) AI Act, despite facing increasing criticism. The French government has consistently maintained a critical stance on this regulation bill concerning artificial intelligence (AI), arguing that certain provisions could hamper innovation and hinder the growth of European AI start-ups. French Economy Minister Bruno Le Maire’s staff emphasized that there are a few points that still appear to us to raise problems.
The Council of EU member-states convened on January 24, wherein French representatives reiterated their concerns. However, the legislative process for the AI Act is entering its final stages, with countries set to vote on its ratification on February 2. France’s reluctance to finalize its position has raised doubts about the outcome of the vote and sparked apprehension within the cultural industries.
Since June 2023, France has consistently contended that specific aspects of the bill could stifle innovation and inhibit the growth of European AI start-ups. Companies like Mistral AI, LightOn, and Hugging Face, which aim to compete with American giants like OpenAI and Google, could be adversely affected by the proposed regulations. Despite the political agreement reached on December 8, 2023, between the EU Parliament, Council, and Commission, Paris has persisted in its criticism.
President Emmanuel Macron expressed skepticism about heavily regulating foundational models more than other countries, citing the potential negative impact on services like the chatbot ChatGPT, which rely on large software programs capable of generating text or images. France’s economy ministry has actively participated in technical meetings to refine the AI Act, but even after the conclusion of these meetings on January 19, Paris has remained on the offensive.
The primary point of contention for France lies in copyright provisions within the bill. French officials assert that while the respect of copyright is crucial, alternative methods should be explored to safeguard it without making the manufacturing secrets of AI models public. Manufacturers are required to publish a sufficiently detailed summary of the data used to train their software. This summary would enable copyright holders to identify if their texts or images have been included and request their removal or negotiate remuneration.
However, French companies like Mistral AI expressed concerns about divulging their manufacturing recipes to competitors. As a result, a provision was introduced to address the protection of trade secrets. The French economy ministry welcomed this change as a step in the right direction but called for further modifications, including limiting the sharing of data summaries to a trusted third party such as the future European AI Office. This office would be responsible for sharing information with copyright holders upon request. France has advocated for this idea since December 2023, without success.
France’s ongoing pressure on the EU’s AI Act reflects its determination to address its concerns and ensure that it offers adequate protection for innovation and the interests of European AI start-ups. As the countries prepare to vote on the ratification of the act, the outcome remains uncertain, and the debate surrounding the regulation of AI continues to unfold.
Overall, while France’s criticisms highlight potential flaws in the AI Act from their perspective, it is important to consider the balance between protecting copyright and promoting innovation. The final decision will shape the future of AI regulation in Europe and determine how the continent competes with global AI powerhouses.
