European Union lawmakers are well on their way to banning the use of remote biometric surveillance for general law enforcement purposes. However, this did not stop MPs in France from voting to introduce artificial intelligence to monitor suspicious behavior in public spaces during the 2024 Paris Olympics.
On Thursday, parliament approved a plan to use automated behavioral monitoring of public spaces during the Games, ignoring objections from around 40 MPs who wrote an open letter condemning the proposal. The vote followed prior approval by the French Senate. (Via Politico.)
The 2024 Olympic Games will be held in Paris between July 26 and August 11.
The EU’s AI law, the future risk-based framework for regulating the use of artificial intelligence, includes a ban the use of ‘real-time’ remote biometric identification systems in public spaces for law enforcement purposes – with exceptions for investigations allowed in the original draft proposal certain potential victims of crime (such as missing children); to prevent “a specific, serious and immediate threat” to life or physical safety or a terrorist attack; or to identify a particular referred perpetrator or suspect of a crime. Although MEPs pushed for a more comprehensive ban.
Critics of the French plan say it goes far beyond the limited law enforcement exceptions allowed in the draft proposal — relying on unproven artificial intelligence to identify something as vague as suspicious behavior.
In a statement, Patrick Breyer, a member of the European Parliament from the Pirate Party, commented on the use of what he called “error-prone” and intrusive technology, saying: “The decision of the French Parliament to allow automated behavioral monitoring in public spaces to search ” abnormal behavior” is creating a new reality of mass surveillance that has never been seen in Europe. I expect the court to invalidate this non-discriminatory surveillance legislation as a violation of our fundamental rights.
“Such suspicious machines will misreport countless citizens, are discriminatory, inculcate conformist behavior and are completely useless in catching criminals, as studies and experience have proven. Step by step, just like in China, social diversity is threatened and our open society is replaced by a conformist consumer society.”
The AI Act was proposed by the European Commission almost two years ago, however remains subject to negotiation by the bloc’s institutions – with the docket debate complicated by divisions and ongoing technological developments such as the rise of general artificial intelligence such as GPT-4 OpenAI (with general purpose AI not explicitly included in the original proposal, with (it is highlighted how rapidly evolving the field of artificial intelligence is and therefore the challenge for regulators to create effective, future-proof frameworks for regulating applications of the technology).
This means that the full scope and details of future pan-European legislation have not yet been agreed. And even in the best case – ie. if the bloc’s lawmakers reach a quick compromise — it still may not go live in time for the Paris Olympics. Still, the French move seems awkward to say the least – suggesting the bloc is headed for a new era of legal friction between national security priorities and the protection of EU fundamental rights.
France is one of several EU member states that has repeatedly refused to comply with EU rules on general and non-discriminatory data retention – as opposed to the activity being essential to national security – despite a number of rulings by the bloc’s top court, in which have found fault with such mass data collection regimes. And future waves of legal challenges over the state’s misuse of powerful AI tools for blanket and indiscriminate surveillance could be quickly over.
Meanwhile, the French government’s plan to blanket the Paris Olympics with AI-powered surveillance could still face a challenge at the country’s constitutional court. Thus, it remains to be seen whether the participants of the 2024 Summer Olympics will face algorithms that will evaluate them behaviorally.
CNIL, France’s data protection watchdog, has turned its attention to artificial intelligence in recent months – in January it set up a dedicated department to work on the technology in preparation for the upcoming EU AI law. So I could be very interested in the government plan. (We’ve reached out to the CNIL with questions about its views on the plan and will update this report if it responds.)