Friday witnessed a long-awaited deal by the European Union on the world’s first comprehensive artificial intelligence rules, which will now give way to legalities on AI.

Negotiators from the European Parliament and the bloc’s 27 member countries overcame controversial points such as generative AI and police use of face recognition surveillance that paved the way to sign a tentative political deal for the Artificial Intelligence Act.

European Commissioner Thierry Breton tweeted: "Deal!" adding, "The EU becomes the very first continent to set clear rules for the use of AI."

Deal!#AIAct pic.twitter.com/UwNoqmEHt5

— Thierry Breton (@ThierryBreton) December 8, 2023

This comes after multiple talks this week, with the first session's duration lasting 22 hours before a second meeting on Friday.

Historic!

The EU becomes the very first continent to set clear rules for the use of AI 🇪🇺

The #AIAct is much more than a rulebook — it's a launchpad for EU startups and researchers to lead the global AI race.

The best is yet to come! 👍 pic.twitter.com/W9rths31MU

— Thierry Breton (@ThierryBreton) December 8, 2023

Daniel Friedlaender, head of the European office of the Computer and Communications Industry Association, a tech industry lobby group, said, "Today’s political deal marks the beginning of important and necessary technical work on crucial details of the AI Act, which are still missing". 

The act will still require a vote by the European Parliament early next year, but Brando Benifei, an Italian lawmaker co-leading the efforts, told AP that the vote is a mere formality given that the deal is done.

No exemptions

Benifei, after being asked if it was inclusive of everything, said via text message, "It’s very very good," adding, "Obviously we had to accept some compromises but overall very good."

Related News

'Israel' to deploy autonomous guns that 'track targets' in West Bank

Intercept exposes contradictions in Google's Project Nimbus claims

The actual law wouldn’t go into full effect until 2025 and includes strict financial penalties for violations of up to 35 million euros ($38 million), equating to 7% of a company’s global turnover.

As of yet, the US, China, UK, and world coalitions like the G7 have given their proposals for AI regulation, but they now have to catch up to Europe.

Doctors and health specialists have already warned that AI development should stop unless it is regulated because it might endanger the health of millions of people and constitute an existential threat to civilization.

Read next: Humans must stay in control of AI: European trade union chief warns

Markets outside Europe are not exempt from the rules implemented on AI firms, "After all, it is not efficient to re-train separate models for different markets."

The legislation regulates AI using a risk-based approach, with requirements corresponding to the amount of danger posed by the system, as well as establishes criteria for suppliers of so-called "foundation models" such as ChatGPT, which have been a major issue for regulators due to how sophisticated they are and worries that even skilled people would be replaced.

The AI Act divides AI applications into four risk categories: unacceptable risk, high risk, limited risk, and low or no risk. Unacceptable risk applications are automatically prohibited and cannot be implemented in the bloc.

'Opting out' of the law

Companies building foundation models or the advanced systems that underpin general-purpose services like ChatGPT and Google’s Bard chatbot will be required to provide technical documentation, comply with EU copyright law, and detail the content applied for training.

Researchers have also previously warned that big tech companies' powerful AI systems could be used against us through online disinformation and manipulation, cyberattacks, or the creation of bioweapons.

AI-powered face recognition surveillance systems are considered to be the riskiest of systems, but negotiators reached a compromise after intensive negotiating since EU lawmakers were planning to fully ban the public use of face scanning and other "remote biometric identification" systems over privacy invasion concerns.

Daniel Leufer, a senior policy analyst at the digital rights group Access Now, said, "Whatever the victories may have been in these final negotiations, the fact remains that huge flaws will remain in this final text."

Leufer is echoing the concerns of many rights groups that claim that major loopholes will remain after its implementation such as lack of protection in border control cases and the option for AI developers to opt-out of having their systems classified as high risk.