Agreement between European Parliament and member states regulates artificial intelligence, social media and search engines

Landmark AI Regulation The world’s first comprehensive law regulating artificial intelligence has been agreed in a landmark agreement between the European Parliament and EU member states after 37 hours of lengthy negotiations.

The agreement was described as “historic” by Thierry Breton, the EU commissioner responsible for various European laws that govern social media and search engines and include major companies such as X, TikTok and Google. Ta.

Prime Minister Breton said 100 people were in the same room for almost three days to get the deal done. He said it was “worth a few hours’ sleep” to complete the “historic” deal.

Landmark AI Regulation Carme Artigas, Spain’s Secretary of State for Artificial Intelligence, who moderated the negotiations, said France and Germany supported the document, but that high-tech companies from those countries should encourage small and medium-sized enterprises to innovate. There have also been reports that the United States is fighting for a more moderate approach to the United States.

The agreement will put the EU in line with the United States, China and the United Kingdom in the race to regulate artificial intelligence and protect citizens from risks, including the potential threats to life that many fear from rapidly evolving technologies. You’ll be ahead of the curve.

Officials provided few details about what exactly the future law, which could take effect as early as 2025, would include.

The political agreement between the European Parliament and EU member states on new laws to regulate AI has been a bitter battle, with disputes over the underlying model, which targets general rather than specific purposes.

But there have also been long negotiations over AI-driven surveillance that police, employers and retailers could use to film people in public in real time and detect mental distress.

The European Parliament has enforced a ban on the use of biometric technologies, including real-time surveillance and emotional recognition, with three exceptions, Breton said.

This means that police can only use this invasive technology in the event of an unexpected threat of terrorist attack, when searching for victims, or when prosecuting serious crimes. Brando Benefei, who led the European Parliament’s four-year battle to regulate AI and co-led the parliamentary negotiating team with Romanian lawmaker Dragos Tudrace, said an “independent authority” would need to give the go-ahead for AI regulation. He also said he had secured a guarantee. “Predictive policing” to prevent police abuse and the presumption of innocence in crimes.

Our aim is to ensure that Europe’s AI ecosystem develops with a human-centered approach that respects fundamental rights and human values, builds trust, and raises awareness about how to make the most of this AI. The goal was to enact a law to ensure that. “This is happening before our eyes,” he said at a post-midnight press conference in Brussels.

Landmark AI Regulation
Empty Assembly Room of European Parliament prepared for meeting, Belgium, Europe.

Mr. Tudorache said, “We tried to deny law enforcement the tools they need to fight crime, the tools they need to fight fraud, the tools they need to give our citizens a safe life.” There has never been.” What we achieved was a ban on AI technologies that determine or pre-determine who will commit a crime. ”

The basis of the agreement is a risk-based tier system in which the highest level of regulation applies to machines that pose a hazard. The greatest risk to health, safety and human rights.

Originally, this was intended to include all systems in which he had a business with more than 10,000 users.

Currently, the highest risk category is defined by the number of computer transactions required to train a machine, known as “floating point operations per second” (flops).

According to sources, there is only one model that falls under this new definition: GPT4.

Legislative regulations still place significant obligations on AI services, including to disclose the data used to train machines for everything from writing newspaper articles to diagnosing cancer. Contains basic rules.

Mr. Tudorache said: “We are the first in the world to introduce real regulations for #AI and the digital world of the future powered by AI, guiding the development and advancement of this technology in a human-centric direction.” said. Previously, the EU had made past mistakes in which giant tech companies like Facebook were allowed to grow into multibillion-dollar companies without any obligation to regulate content on their platforms, including election interference. He said he was determined not to. Child abuse and hate speech.

Strong and comprehensive regulation by the EU “could set a strong example for many governments considering regulation,” says
Professor at Columbia Law School and expert on the EU and digital regulation. Anu Bradford said. Other countries “may not emulate all of the provisions, but are likely to emulate many aspects of them.”

AI companies required to comply with EU rules are likely to extend some of their obligations to markets outside the continent, Bradford told The Associated Press. “At the end of the day, retraining separate models for different markets is not efficient,” she said.