Moves to ban live “Big Brother” real time facial recognition technology from being deployed across the streets of the EU or by border officials will be tested in a key vote at the European parliament on Thursday.
The amendment is part of a package of proposals for the world’s first artificial intelligence laws, which could result in firms being fined up to €10m (£8.7m) or removed from trading within the EU for breaches of the rules.
It is contained in one of 12 groups of compromise amendments agreed by a committee of MEPs, whittled down from more than 3,000 submitted a year ago.
But the ban, contained in a final text to be voted on in parliament on Thursday, is expected to be challenged by a group of centre-right MEPs on the grounds that biometric scanning should be deployed to combat serious crime such as terrorism.
If passed the law will also ban “emotional recognition” AI which could be used by employers or police to identify tired workers or drivers.
Charities have expressed concern that live real-time facial recognition would be open to abuse by state agencies and border police.
But Dragos Tudorache, co-rapporteur of the AI Act in the European parliament said he hoped there would be strong support for it to be forbidden.
“There is no stronger safeguard [than this ban]. A border crossing point is a public space. According to the text we have right now, you will not be able to deploy AI biometric recognition technology in a public space,” he said.
The act will also force those generating artificial intelligence to be transparent about which original literature, science research, music and other copyrighted materials it uses to train machine learners.
This will enable bands, academics and others to sue if they think copyright law has been breached.
Co-rapporteur Brando Benifei said he hoped the law would allay concerns over artificial intelligence disrupting employment markets and a potential deluge of fake news, disinformation and interference with human rights.
“With our text, we are also showing what kind of society we want, a society where social storing, predictive policing, biometric categorisation, emotional recognition, and discriminated scraping of facial images from the internet are considered unacceptable practices,” he told reporters.
The amended text of the AI Act will go before the wider parliament in the middle of June and if voted through will represent a “strong” mandate in further discussions with the European Commission, and the Council of the European Union.
The law is expected to be passed by the end of the year.
Many believe the AI Act will become the gold standard of regulation around the globe, adopted by giants such as Google, Microsoft and social media companies.
“Is it known as the Brussels effect. If the EU moves first and has sensible standards other countries will start with the EU rules when designing their own regulation,” said Zach Meyers, research fellow at the Centre for European Reform.
Meyers added: “Even if they don’t, companies may voluntarily adopt the EU rules globally because it makes the cost of doing business cheaper.”
Kim van Sparrentak, the Dutch Green party MEP, said the use of live scanning, made possible by AI, was “completely against our fundamental rights” and “an unacceptable risk”.
The AI Act, which is the first of its kind, has been in the making for almost two years, with fresh amendments added recently to address risks posed by “general purpose” AI systems, including ChatGPT.
Asked if the EU was not acting too late to address ChatGPT just a month before the entire European parliament is asked to vote on the AI Act, Tudorache said: “If we are late, where are all the other jurisdictions that haven’t even started to consider regulation?”
