Financial News Agency, December 12 (Editor Xia Junxiong) On Monday (December 11) local time, French President Emmanuel Macron warned that compared to competitors, the European Union's artificial intelligence (AI) law may hinder the development of European technology companies.

The European Union has reached an agreement on the Artificial Intelligence Act

On Friday, after lengthy negotiations, EU member states, the European Parliament, and the European Commission finally reached a political agreement on the Artificial Intelligence Act, one step closer to comprehensive AI legislation.

According to the European Parliament's statement, the agreement reached by lawmakers includes banning several AI applications, such as untargeted scraping of images to create facial recognition libraries, and setting rules for systems determined by lawmakers to be high-risk. The agreement also includes transparency rules for generic AI systems and the models that drive these systems.

Companies that fail to comply with the new regulations will face fines of up to 7% of global turnover.

The Artificial Intelligence Act still needs to be approved by EU member states in the next few weeks. If it is officially approved, the bill is expected to become the world's first comprehensive regulation on AI. The Artificial Intelligence Act is likely to come into effect as early as 2025.

France is dissatisfied with the Artificial Intelligence Act

Speaking in the southwestern French city of Toulouse on Monday, Macron criticized the Artificial Intelligence Act. He said, “We can decide to regulate faster and more vigorously than our main competitors, but we're going to regulate something we're no longer producing or inventing, which is never a good idea.”

Macron said he is concerned that the Artificial Intelligence Act will impose the strictest regulation on basic models in the world. The basic model is the technology that supports generative AI products such as ChatGPT, which can generate large amounts of human-like text, images, and code within seconds.

Previously, regulation of the underlying model was the most contentious part of the Artificial Intelligence Act negotiations. Some experts and legislators have proposed a hierarchical approach to regulating basic models. ChatGPT is defined as a very capable foundational model. Such foundational models are subject to additional obligations, including regular reviews to uncover potential loopholes.

Some legislators believe that the smaller models may have the same level of risk. However, countries such as France tend to let companies that develop generative AI models regulate themselves rather than set strict rules.

Macron said, “France is probably the largest country in the AI field in mainland Europe, and is on par with the UK, and the UK will not have such regulations on the basic model. But most importantly, we are all far behind China and the US.”

Macron's statement on the Artificial Intelligence Act is critical, which means the bill is still facing resistance. It is reported that France, Germany, and Italy are in preliminary discussions on seeking amendments or blocking the passage of the bill.

People familiar with the negotiations said they believed an agreement would be reached in the end despite objections.