EU takes world’s first step in regulating AI

The EU took the initial step on Wednesday to start regulating artificial intelligence, in a first global attempt to manage risks from this rapidly developing technology.

The European Parliament overwhelmingly approved a draft law known as the AI Act, which puts restrictions on the technology’s riskiest uses.

Approved with 499 votes in favour, 28 against and 93 abstentions, the draft law will form the basis for discussions between parliament, the Council and the European Commission.

European Parliament President Roberta Metsola said after the vote the intention is to have a finalised law by the end of the year.

The AI Act would ban the real time use of facial recognition software to avoid the risk of mass surveillance and require makers of AI systems like ChatGPT to disclose more about the data used to create their programmes. Another provision is a requirement for AI generated material to be identified as so.

The legislation is likely to be a trailblazer internationally amid global concerns on the dangers posed by AI and its impact on jobs and society.

The EU has been debating the technology for two years but the issue became more urgent after last year’s release of ChatGPT which intensified concerns about AI.

Addressing a press conference after the vote, Metsola said the AI Act approved by parliament provided a “balanced and human-centred” approach to what she described as the first rules governing AI.

“These are the first regulations to manage risk and provide for the lawful use of AI and are consistent with our will to be innovators in full respect of the values that underpin the union,” she said.

Metsola, who started the press conference by saying that she asked ChatGPT what to say, insisted that “emotions, will and judgement cannot be artificialized”.

She added: “These [qualities] belong to us, here… and we do it responsibly.”

The co-rapporteurs of the report that parliament voted on, Brando Benifei and Dragoş Tudorache said the challenge was trying to find a balance.

“We had to identify the risks without overburdening creators and users of AI but we had to ban unacceptable uses of the technology that go against our fundamental values,” Benifei said.

Dragoş said it was important that regulation is accompanied by education and reskilling of the workforce. “We have to invest more in education and reskilling because the challenges posed by AI will create forces who will want to resist this technology.”