European Union now has a law that regulates the use of AI

The plenary session of European Parliament approved this Wednesday an ambitious regulation on the use of artificial intelligence in the EU, in the last step before its formal adoption throughout the bloc.

In the plenary session this Wednesday, the law obtained broad support of 523 votes in favor, 46 against and 49 abstentions.

The original bill, considered unprecedented worldwide, had been presented by the European Commission, the executive arm of the European Union, in April 2021.

However, the appearance at the end of 2022 of ChatGPT, from the Californian company OpenAI, capable of writing dissertations, poems or making translations in seconds, gave the project a new dimension.

This system revealed the enormous potential of AI, but also exposed its risks. The spread of fake photographs or videos highlighted the danger of manipulation of public opinion.

“This is a historic day in our long path towards the regulation of Artificial Intelligence,” said Italian legislator Brando Benifei, one of the speakers of the text, along with Romanian Dragos Tudorache.

In his last speech before the vote, Benifei said that it is about “first regulation in the world that defines a clear path towards the development of Artificial Intelligence focused on human beings.”

In turn, Tudorache said that “We have achieved a very delicate balance between the interest in innovation and the interests that must be protected.”

For his part, the European Commissioner for the Internal Market, Thierry Breton, welcomed that this Wednesday’s vote in the European Parliament means that the EU is a “standard setter” regarding Artificial Intelligence “reliable”.

After this Wednesday’s vote, it is expected that the 27 countries of the bloc will support this law in April, and that the text will be published in the Official Journal of the EU in May.

The veto on the use of forms of AI considered high risk will apply at the end of this year, and the rules for so-called generative AI, such as ChatGPT, will apply 12 months after the standard is ratified.

Models with different requirements

The adopted regulations provide that AI models of “general purpose” They must comply with transparency obligations, as well as European copyright regulations.

Meanwhile, the systems considered “high-risk” (such as those used in critical infrastructure, education or law enforcement) will be subject to more stringent requirements.

These systems will have to provide for the implementation of a mandatory impact analysis on fundamental rights.

The text supported by MEPs also plans to prohibit citizen rating or mass surveillance systems, or even the remote biometric identification of people in public places.

The legislation defines means of surveillance and sanctions, through the creation of a European AI Office.

This body may impose fines ranging from 7.5 million euros or 1.5% of the turnover of the affected companies, up to 35 million euros or the 7% of overall billing, depending on the violation and the size of the company.

This regulation determines that citizens must be clearly informed that they are dealing with artificial intelligence resources.

The law also prohibits police officers from using real-time facial recognition technology, with exceptions for law enforcement if they are searching for a person convicted of or suspected of a serious crime, such as rape or terrorism.

The police may request the use of the technology to find victims of kidnapping or human trafficking, but with approval from a judge or other judicial authority, and for use limited in time and place.

It may interest you