The European Parliament’s man on artificial intelligence wants to introduce tougher restrictions on controversial facial recognition tools in Europe.
Brando Benifei, an Italian from the centre-left Socialists and Democrats group, has been named as one of the Parliament’s leads to amend the Artificial Intelligence Act — the world’s first law to ban or restrict AI that poses a risk to humans.
Benifei wants to put stricter controls on the bloc’s burgeoning biometrics industry — which is developing tech to track everything we do — and ensure rules are implemented consistently across Europe.
“We’re probably going to intervene to strengthen the provisions in the [draft] text on biometric surveillance,” the lawmaker said in an interview. “We will propose modifications so it is clearer that we won’t allow any national interpretations on this high risk tech.”
The European Commission, the bloc’s executive body, unveiled regulations to govern the advancing AI field in April, with proposals to set limits around the use of the tech in a range of activities, from hiring decisions to bank lending, school enrolment selections and exam scoring.
Getting ahead of a tech field dominated by China and the US is a tall order for officials.
‘High risk’ AI
Some uses of AI would be banned altogether under the draft rules, including 'real-time' facial recognition in public spaces, though there would be several exemptions for national security and other purposes, including the search for missing children.
Several analysts say the rules, as currently written, would not prevent the sale of European-made surveillance tools to an authoritarian regime like China — a potential loophole Benifei and colleagues say they will study closely.
According to the Italian, “The security exceptions in the draft are too generic. We don’t want to leave a backdoor open to authoritarian regimes. We need to be careful we don’t underestimate the geopolitical factors here,” he added.
But getting ahead of a tech field dominated by China and the US is a tall order for European officials. Benifei will be negotiating for the Parliament with the Council of the EU, which represents the 27 EU countries — a process that will likely take several years.
Facial scanning continues to creep into use in Europe, operating with little oversight.
‘End of anonymity’
In the meantime, facial scanning — which combines surveillance cameras, computer vision and predictive imaging — continues to creep into use in Europe, operating with little oversight. Police forces in almost all EU countries already use face recognition tech or plan to introduce it.
In June, top privacy watchdogs, the European Data Protection Supervisor, which acts as the EU’s independent data authority, and the European Data Protection Board, which helps countries implement GDPR, called for a total ban on using AI to automatically recognise people.
“Deploying remote biometric identification in publicly accessible spaces means the end of anonymity in those places,” the heads of the two bodies, Andrea Jelinek and Wojciech Wiewiórowski, said in a joint statement.
Privacy activists argue that face scan technology is potentially authoritarian, because it captures images without consent. They also raise concerns about the potential for racial bias. If an AI system is trained primarily on white male faces, but fewer women and people of colour, it will be less accurate for the latter groups.
Less accuracy means more misidentifications. An investigation by the UK civil liberties group Big Brother Watch in 2018 found that the automated system used by the Metropolitan Police Service in London had a false positive rate of 98%, and that the police retained images of thousands of innocent citizens, for future searches.
I’m concerned that we don’t stop innovation.
Innovation killer?
Supporters of regulation say human oversight is needed for a rapidly developing field that presents new risks to individual privacy and livelihoods. Others warn that the new rules could stifle creation with lasting economic consequences for Europe.
“I’m concerned that we don’t stop innovation,” said Benifei. “I support the idea of [regulatory] sandboxes.”
The sandbox concept, which entails a collaborative approach between regulators and companies, has become a popular vehicle for testing new technologies in several European countries, including the UK, Norway and France.
In Norway’s AI sandbox, for example, a company called Secure Practice is exploring the legality of using AI to profile individuals. The startup, based in Trondheim, is seeing if it can infer employee training needs based on an AI-read of personality types.
Another sandbox participant, a startup called Age Labs, is experimenting with an AI tool to predict a person’s likelihood of disease based on a blood sample.
Major tech companies are wary of the EU’s plans to regulate.
Who controls AI?
“I’m convinced AI can make life better for people; it can make a lot of things easier and [help us] develop more economic opportunities,” Benifei added.
But the Italian politician says he also sees inherent risks in one of the world’s most transformative technologies. “What scares me is control — if AI systems lack transparency, then there are big risks for citizens,” he said.
The major tech companies are wary of the EU’s plans to regulate. Civil society groups, on the other hand, fear the rules won’t go far enough.
The various lobbying efforts are already piling high in Benifei’s inbox. “Everyone is contacting me from all sides,” he said.
