European Union Passes World’s First Comprehensive AI Regulations

  • Oops!
    Something went wrong.
    Please try again later.

The European Union passed first-of-its-kind, comprehensive artificial intelligence legislation Wednesday.

The EU AI Act has been the subject of discussion in Parliament for several years after first being introduced in 2021, ahead of the eventual public-facing zeitgeist around the technology—and specifically generative AI—brought on by OpenAI in 2022.

More from Sourcing Journal

Parliament members largely endorsed the legislation, with 523 members voting in favor of the act and 46 members voting against the legislation. 46 members did not cast votes.

With that approval, the legislation is set to take effect later this year.

The President of the European Parliament, Roberta Metsola, called the newly adopted legislation “trailblazing” in a social media post.

Despite the fact that it has historically lagged on the development of technologies, especially as compared with the U.S. and China, Metsola said in a speech that the legislation will set the EU ahead.

“Our groundbreaking AI law will allow us to be world leaders in digital and tech innovation, based on EU democratic values,” Metsola said. “Europe has the ability to set the tone worldwide and lead responsibly.”

Deirdre Clune, a member of the European Parliament for Ireland South, served as lead lawmaker for the drafting of the EU AI Act. When the European Parliament expressed its support for the regulation in December, Clune said it would serve as a major step forward in protecting EU citizens.

“By guaranteeing the safety and fundamental rights of people and businesses, the Act will support the human-centric, transparent and responsible development, deployment and take-up of AI in the EU,” Clune said in a statement. “We must find a balance between regulating and promoting new technologies.”

Experts said they expect that the legislation will serve as a starting point for other countries looking to regulate AI, which is developing rapidly—and, in many places, without clear guidelines.

In the United States, several states have proposed AI-centered legislation, but so far, nothing has come forward at the national level. In October, President Biden signed an Executive Order that could in time influence national legislation in the U.S.

Nitish Mittal, partner at Everest Group, said he expects the effects of the EU AI Act to be far reaching. While some have compared the potential halo effect from the newly passed legislation to that of the GDPR, adopted by the EU in 2016, Mittal said he believes this legislation will prove more influential globally.

“I believe the impact of AI is going to be much more broad, and that’s [why] I think that the EU AI Act might have a knock-on effect on this conversation [globally], because this is something most countries need,” he said.

Because of the rate at which AI has changed since the first proposal of regulation around the technology, the bill has at times shifted to include updates like generative AI. Mittal said he anticipates that the newly adopted regulations will need some wiggle room to stay relevant in a constantly changing technology landscape.

“Any policy around this, if it claims to be ironclad, I think is not practical because this is a market where things that we know on Monday can be outdated by Thursday, just given the pace of development,” he explained. “It’d not be wise for any government—and even any company—to think that they know everything about AI…This needs to be a series of smaller reviews and milestones.”

The law assesses use cases for and development of the technology on a risk-based scale, mandating certain provisions for high-risk systems and use cases and banning the technology in places that the EU has deemed “unacceptable risk” is present.

Some stakeholders have expressed concerns about implementation and enforcement of the AI Act, which will begin six months after the new regulations make it into the EU law books. That entry is expected in May or June.

For the best chance of success with implementation, Mittal said, both the at-large EU and the individual participating countries will need to show a willingness to collaborate with companies as they navigate the incoming requirements.

“Interpretation will even out as we get into the weeds of this, because the EU doesn’t want to act as a inhibitor to AI innovation, and because the pace of innovation is so rapid,” Mittal said.

He also suggested that brands and retailers band together to understand the scope of the legislation, especially because the regulations affect companies doing business in the EU, not just companies headquartered in the EU.

“I don’t think that you’re in this alone. If [companies] can ask their peers in the industry, the partners that they work with, this is going to be a space where collaboration and conversation is essential. I don’t think anybody’s figured this out,” he said. “Cooperation and collaboration versus competition is going to be really important.”

If the implementation of the law goes smoothly for companies and governments alike, Mittal said he expects the EU AI Act will serve as a reminder that proposing and passing legislation in line with a peak in a technology’s hype cycle is possible with a bit of foresight.

“This is really a watershed moment for AI, but also [for] how we deal with technology revolutions in the coming years,” Mittal told Sourcing Journal. “As we get into this world of…innovation happening much more rapidly, the learnings, the muscle memory that we gained from reacting so rapidly to this is something that will hopefully serve as well for future technology waves—not just for this one.”