The EU’s law on artificial intelligence (AI) has come into force. Member states now have two years to implement the requirements into national law. The law will regulate AI more strictly and uniformly in the European Union. It is intended to better protect fundamental rights, democracy and the rule of law when dealing with the technology. What the law means exactly:

What is artificial intelligence anyway?

Artificial intelligence usually refers to applications based on machine learning, in which software searches through large amounts of data for similarities and draws conclusions from them. This allows human abilities such as logical thinking, learning, planning and creativity to be imitated. This enables machines to perceive and react to their environment, for example.

AI is already being used in many areas. For example, such programs can evaluate CT scans faster and with greater accuracy than humans. Self-driving cars, in turn, try to predict the behavior of other road users. And chatbots or automatic playlists from streaming services also work with AI.

Why is such a law needed in the EU?

The law aims to make the use of AI safer in the European Union. It is intended to ensure that AI systems are as transparent, traceable, non-discriminatory and environmentally friendly as possible. An important aspect is that the AI ​​systems are monitored by people and not just by other technologies.

What rules does the law contain?

The regulations provide for AI applications to be classified into different risk groups. Systems that are considered particularly high-risk and are used, for example, in critical infrastructure or in education and healthcare must meet strict requirements. Applications with a lower risk are subject to fewer obligations.

AI applications that violate EU values ​​are also completely banned. This includes the evaluation of social behavior (“social scoring”). This is used in China to classify citizens into behavioral categories.

What does it mean for consumers?

The law is intended to better protect consumers from risky AI applications. Facial recognition in public spaces – for example through video surveillance in public places – is generally not permitted. Emotion recognition in the workplace and in educational institutions is also prohibited in the EU under the law.

In addition, AI applications must be labelled more transparently. This will make it easier for consumers to identify which programs use artificial intelligence. Private individuals who discover violations of the regulations can complain to national authorities.

What exactly will change from August 1st?

Not much at first. The AI ​​law is intended to take effect gradually. Some regulations must then be implemented promptly by the member states – such as the ban on AI systems that pose “unacceptable risks”. These are systems that are classified as a threat to people. Their ban applies after just six months.

A code of conduct for providers of AI models is to be completed by April next year, the EU Commission announced before the law came into force.

After two years, most of the points in the law must be fully implemented. High-risk systems, however, will have more time to meet the requirements. The obligations that apply to them will apply after three years.

What happens if someone doesn’t follow the rules?

Violations can result in severe penalties: for example, if prohibited technology is used, up to 35 million euros or – in the case of companies – up to seven percent of the worldwide annual turnover of the previous financial year. However, the exact level of punishment must be determined by the countries, as the Commission announced.

Other violations of the law may result in fines of up to 15 million euros or, in the case of companies, up to three percent of the worldwide annual turnover of the previous financial year.

Is there criticism of the law?

Experts have recently been discussing whether the law will give AI a boost or whether its development will perhaps even be slowed down. In the end, this probably also depends on the respective national implementation. Green Party member of the Bundestag Tobias Bacherle urged that a regulation for AI-supported biometric surveillance in Germany must now be found. In the wrong hands, it could easily be misused to undermine civil liberties, he said.

Federal Digital Minister Volker Wissing believes that some of the requirements of the EU law go too far. “I would have liked to see more innovation-friendly regulation,” the FDP politician told the German Press Agency. “But in the end, it has to be a compromise, which is better than no regulation.” Now, the focus in Germany is on implementing the law with as little bureaucracy as possible.

In addition, it was previously criticized that many requirements could soon become outdated again given the rapid technical development of AI applications and the gradual implementation of the regulations. The Commission announced that it would carry out an annual review to determine whether the list of “high-risk” applications needed to be revised or expanded.