On Tuesday, the European Union officially approved a critical AI regulation reached in December last year. This “landmark” AI regulation standard and regulation for global AI regulation will come into effect next month, setting a potential benchmark for the application and popularization of artificial intelligence in human society, commerce and daily life.

On a regulatory level, the EU Artificial Intelligence Regulation Act is more comprehensive than the light-touch voluntary compliance approach proposed by the US. Two months ago, EU lawmakers supported artificial intelligence regulation legislation drafted by the European Commission starting in 2021, after the European Commission made some key changes to the legislation.

In recent months, as many generative artificial intelligence systems such as the epoch-making AI chatbot ChatGPT developed by OpenAI and the AI chatbot Gemini launched by Google (Google) have become more and more popular with global consumers, people's concerns that artificial intelligence fuels misinformation, fake news, and copyrighted materials have spread more and more around the world, increasingly showing the importance of AI regulatory guidelines

Belgian Minister of Digitalization Matthew Michel said in a statement: “This landmark law is the first formal law of its kind in the world. It can effectively address global technological challenges and create new opportunities for our society and economy.”

He added: “With this officially passed artificial intelligence regulation, European countries have emphasized the importance of trust, transparency and accountability in dealing with this new technology, while ensuring that this rapidly changing technology can flourish and promote innovation in Europe.”

According to information, the EU's Artificial Intelligence Act imposes strict transparency obligations on high-risk artificial intelligence systems, and the requirements for general-purpose artificial intelligence models will be relatively light.

The bill also limits the government's use of real-time biometric surveillance in public places to certain criminal cases to prevent terrorist attacks and search for groups suspected of the most serious crimes.

Patrick van Eecke (Patrick van Eecke) of Cooley Law Office (Cooley) said the impact of the new legislation will go far beyond the EU's Group of 27 and spread to every corner of the world.

“This bill will have a global impact. In particular, companies outside the EU that use EU customer data in their artificial intelligence platforms will need to comply with this regulation. Other countries and regions may use the EU Artificial Intelligence Act as a benchmark blueprint, just as they did with the General Data Protection Regulation,” he said, referring to the EU's data privacy regulation benchmark rules.

Relevant AI regulatory bans and some regulations will be gradually implemented within the first 6 months after the regulations officially come into effect. This means that although full implementation may take until 2026, some very critical AI regulations will begin to be implemented quickly within a short period of time after the regulations come into effect, thereby rapidly affecting AI usage in related fields.

Therefore, although the new regulations will be the first to be fully popularized and will eventually be fully implemented in 2026, the ban on the use of artificial intelligence in social scoring, predictive police management, and the untargeted capture of facial images from the Internet or CCTV footage will soon take effect within 6 months. In other words, some AI regulations in the EU will take effect within 6 months, which means that the legal status of the regulations themselves will begin to be gradually established within 6 months, but the specific implementation details and execution time of AI regulations vary according to different regulations.

The rules also indicate that the relevant regulatory obligations for general artificial intelligence models will take effect after 12 months, and the regulatory rules for deep artificial intelligence systems embedded in regulated products will take effect after 36 months.

In terms of penalties for violations of the EU's AI regulatory system, fines for violations range from 7.5 million euros (about 8.2 million US dollars) or 1.5% of total turnover to 35 million euros, or even 7% of global turnover, depending on the type of violation.