The European Union (EU) accepted the world’s first artificial intelligence law, which imposes strict regulations on the use of technologies in this field.

The law, adopted by EU member states on May 25, came into force today after its publication in the Official Gazette of the Union on July 12.

The rules, known as the world’s first “Artificial Intelligence Law”, will gradually harmonise the regulations in this field in 27 EU member countries.

Essentially, the aim is for the artificial intelligence systems used in EU countries to operate in a way that does not harm fundamental human rights and security.

Therefore, the rules group artificial intelligence systems according to the harm they cause to society. As the degree of risk increases, the rules become stricter.

For example, areas of the systems that could seriously affect human life and threaten health, such as transportation, exam scoring, surgical procedures, employment procedures, evaluations for bank loans, and examination of visa applications, were classified as “high-risk”.

Low-risk areas include applications such as artificial intelligence-supported video games or spam filters.

The EU Commission states that most systems currently used in Union member countries fall into this category.

Conditions for using artificial intelligence applications are determined according to the degree of risk, and the responsibilities of the application’s distributors, providers, and users are clearly defined.

Systems that fall into the “unacceptable” risk group are entirely banned. The ban includes taking facial images from the internet or closed camera systems and performing biometric classification in the workplace and educational institutions to extract sensitive data such as emotion recognition, social scoring, sexual orientation or religious beliefs.

Artificial intelligence systems that manipulate human behaviour or exploit human vulnerabilities are also banned.

Special rules apply to large systems that can perform various tasks, such as creating video, text, and images, speaking another language, calculating, or writing computer code.

It is envisaged to ensure that these “general purpose artificial intelligence” systems comply with various transparency criteria before being released to the market.

Companies that do not comply with the rules face fines of millions of euros proportional to their global turnover.

For example, if a banned technology is used, the company could be fined up to 35 million euros or 7 percent of its global turnover. In case of violation of other provisions, fines of up to 15 million euros or 3 percent of turnover may be imposed.

The bans will become applicable six months after the law comes into force today.

Governance rules and obligations for general-purpose AI models will come into force after 12 months.

The law will be applicable in all its dimensions after two years.