MEPs to vote on proposed ban on ‘Big Brother’ AI facial recognition on streets
Moves to ban live “Big Brother” facial recognition technology from being deployed across the streets of the EU 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 see firms fined up to €10m (£8.7m) or removed from trading within the EU for breaches of the rules.
But the ban 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.
An amendment to Article 5 of the proposed Artificial Intelligence Act prohibiting the use of cameras to follow people around shops, streets, parks or any other public places will go before the committee on Thursday. It will also ban companies from using AI to identify individuals by match scans on CCTV with images posted by users on Instagram, Facebook or other social media platforms.
Another amendment bans the use of so-called “emotional recognition”, which might allow computer systems to recognise when someone is tired, amid concern it could be misused by employers or the state.
Such surveillance is already used in parts of China by traffic police. A BBC documentary on its deployment in China found it was not foolproof, with one truck driver accused of using a phone while driving who, on closer inspection, was found to have been scratching his nose.
A proposed ban on live biometric scanning is one of 12 so-called compromise amendments, whittled down from more than 3,000, to the AI Act.
Sources say it is the only one not expected to pass at the committee stage on Thursday.
“It was a point of division between groups. There will be a number of colleagues in the EPP (European People’s party group) who will very likely vote against that part of the compromise,” Dragos Tudorache, co-rapporteur of the AI Act told the Guardian.
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.
Tudorache said he is confident the AI Act will calm anxieties among the general public that have become heightened over the past month amid talk of AI destroying jobs, taking over from medics, or being used by bad actors to interfere with elections and spread disinformation.
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, he said: “If we are late, where are all the other jurisdictions that haven’t even started to consider regulation?”
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.”