European Union Parliament adopts ‘groundbreaking’ regulations on artificial intelligence with the passing of AI Act

On Wednesday, the European Parliament officially approved the most comprehensive regulations in the world for governing artificial intelligence, which includes advanced systems such as OpenAI’s ChatGPT.

The AI law focuses on riskier uses of the technology in the private and public sectors, stricter obligations for providers, stricter transparency rules for the most effective models such as ChatGPT, and an outright ban on tools deemed too dangerous.

Senior European Union officials say the rules proposed in 2021 will protect citizens from the risks of rapidly developing technology while boosting innovation on the continent.

EU chief Ursula von der Leyen hailed the vote, which created “a ground-breaking framework for innovative AI with clear safeguards”.

“This will benefit Europe’s fantastic talent pool. And create a model for reliable AI across the world,” he said at X.

The text was approved by 523 EU legislators in favor and 46 against. The 27 EU countries are expected to approve the law in April before it is published in the EU’s Official Journal in May or June.

Brussels has been scrambling to adopt new rules since OpenAI’s Microsoft-backed ChatGPT arrived on the scene in late 2022 and unleashed a global AI race.

There was excitement around generative AI when ChatGPT amazed the world with its human-like capabilities – from digesting complex text to producing poems in seconds or passing medical tests.

Other examples include DALL-E and Midjourney, which generate images, while others generate sounds using simple everyday language.

But with the excitement came threats quickly realized – not least that artificial intelligence-generated deep fakes of audio and video would turbocharge disinformation campaigns.

“Today is another historic day on our long road to AI regulation,” said Brando Benifei, an Italian lawmaker who pushed the text through parliament with Romanian MEP Dragos Tudorache.

“We managed to strike a very delicate balance between innovation and the conservation interest,” Tudorache told reporters before the vote.

Rules for AI models such as ChatGPT will come into effect 12 months after the law is formalized, while companies must comply with most other regulations after two years.

– AI Police Limitations –

The EU rules, known as the artificial intelligence loan, have a risk-based approach: the riskier the system, the stricter the requirements – and outright bans on artificial intelligence tools that pose the biggest threat.

For example, providers of high-risk AI must conduct risk assessments and ensure that their products comply with the law before making them available to the public.

“We will regulate as little as possible and as much as necessary, in a proportionate way, for AI models,” EU Internal Market Commissioner Thierry Breton said.

Violations can result in fines ranging from 7.5 million to 35 million euros ($8.2 to $38.2 million) depending on the type of violation and the size of the company.

The use of artificial intelligence for predictive policing and systems that use biometric data to infer an individual’s race, religion or sexual orientation are strictly prohibited.

The rules also prohibit real-time facial recognition in public places, with some exceptions for law enforcement, though police must seek permission from a judicial authority before deploying AI.

Digital civil rights group Access Now said the bans did not go far enough.

“The final text is riddled with loopholes, exemptions and exceptions, which means it fails to protect people and their human rights from some of the most dangerous uses of AI,” it said in a statement.

– The EU “resisted the pressure” –

As AI is likely to change every aspect of Europeans’ lives, and as major tech companies compete for dominance in lucrative markets, the EU has been the target of intense lobbying.

Watchdogs pointed out on Tuesday the campaigning of French artificial intelligence startup Mistral AI and Germany’s Aleph Alpha, as well as US tech giants such as Google and Microsoft.

They warned that corporate lobbying could further undermine the law’s implementation.

“Many details of the AI law are still open and need to be clarified in numerous implementing acts, for example regarding standards, thresholds or transparency obligations,” said three watchdogs based in Belgium, France and Germany.

Legislator Tudorache said the bill was “one of the … most heavily lobbied pieces of legislation, certainly in this mandate,” but stressed, “We resisted the pressure.”

Organizations representing Europe’s creative and cultural sectors welcomed the vote in a joint statement, but called on the EU to ensure that “these important rules are put into practice in a meaningful and effective way”.