Law on intelligence – Kommersant

The profile committees of the European Parliament have previously approved the Law on Artificial Intelligence, which regulates various aspects of the use of systems and tools based on AI. In particular, the draft contains a ban on the use of face recognition technology in public places, the use of predictive analytics by the police, and measures to regulate generative AI systems. According to experts, the law, if adopted, will affect not only the European, but also the global market.

On Thursday, two committees of the European Parliament – on the internal market and on civil liberties – approved the draft law (.pdf) on artificial intelligence (AI Act). 84 deputies voted for, 7 – against, 12 abstained. In June, the European Parliament as a whole will have to vote. After that, the document will be discussed with the European Commission, as well as with EU member states. It is expected that the law will be finally adopted at the end of this year or early next year.

• The 144-page bill outlines rules for the use of various AI-based systems, including facial recognition systems, predictive analytics, generative AI, self-driving cars, and others.
• All AI-based systems and tools under the new law must be classified by risk level – from low to unacceptable.

The law will become the world’s first example of a versatile and comprehensive regulation of the use of artificial intelligence, which in the future will in one way or another be guided by the authorities of other countries and corporations. “We are on the verge of creating legislation that is truly significant for the digital environment, not only for Europe, but for the whole world,” said Brando Benifei, one of the authors of the bill.

This is due, among other things, to the fact that companies are likely to focus on EU standards when developing AI systems. As Oxford Internet Institute lecturer and AI regulation expert Johann Lauks noted, “Europeans are quite rich and there are many of them,” so developers will be guided by EU regulations rather than developing different systems for different markets.

The bill prohibits the use of a number of tools and systems that allow using AI to monitor citizens and, in particular, create profiles of potential criminals.

Deputies from the left and center-left parties, as well as many human rights organizations, insisted on a ban on such practices. The centre-right European People’s Party, by contrast, opposed the measure.

“With its ban list, Parliament … is signaling to governments and developers that some uses of AI are too harmful to be allowed,” said Sarah Chander, Senior Policy Adviser at European Digital Rights (EDRi).

Among other things, the AI ​​Law provides for a ban on the use of facial recognition systems and other “remote biometric identification” systems in real time in public places. In recent years, they have become widespread in different countries.

In addition to face recognition in the classical sense, which allows you to determine the identity of a person from video, the law also forbids so-called emotion recognition. It works in a similar way and reveals a person’s condition – this can be used by employers or the police to identify tired workers or drivers.

The law also prohibits the use of predictive analytics systems by the police for crime prevention. Such systems, analyzing huge amounts of data, make a prediction about the likelihood of a crime and compile profiles of potential offenders. In addition to the general opacity of such forecasting, such AI systems are already criticized for the fact that they reproduce the stereotypes that prevail in society. For example, members of low-income populations can be classified by AI as a high-risk group only because they are more likely to be in contact with social services and their data is more likely to be found in government databases, on the basis of which AI makes predictions.

The law contains norms for strict regulation of various “high-risk systems based on AI.” These include physical objects – unmanned vehicles, AI-enabled medical devices, AI-based critical infrastructure control systems, etc. Failure in their operation threatens lives of people or leads to other serious consequences.

As far as generative AI is concerned, that is, in particular, AI-based chatbots such as ChatGPT, the law does not prohibit their use, but rather tightly regulates it. Among other things, the creators of such systems must disclose information about which copyrighted books, music, paintings, etc. were used in AI training. This should help fight copyright infringement and plagiarism.

In general, companies using various AI tools will need to assess their potential risks in terms of “health, safety, fundamental rights, the environment, democracy, and the rule of law.”

Under the law, violations of these regulations could result in a fine of up to €30 million, or 6% of the company’s annual global revenue. True, the first two years after the entry into force of the law, it will operate in a test mode. To enforce the law, an EU AI Office will be set up to oversee its implementation together with regulators in individual EU countries.

Yana Rozhdestvenskaya