Italy has introduced a new ban on facial recognition, but it includes a blatant exemption that will reverberate across the EU.

The new ruling prohibits the use of facial recognition technology (FRT) — unless it is used for crime fighting or judicial investigation. Data protection of Italy Agency pledged to maintain the ban until new legislation is passed, or at least until the end of 2023.

“The moratorium stems from the need to regulate the eligibility requirements, conditions and guarantees related to facial recognition, in accordance with the principle of proportionality,” the watchdog said in Monday. a statement.

Greetings, humanoids

Subscribe to our newsletter now to get a weekly roundup of our favorite AI stories delivered to your inbox.

Civil rights advocates cautiously welcomed the new restrictions, but expressed concern that the loopholes could be manipulated.

“This would affect all private use. Obviously some big questions about how broadly ‘fighting crime’ will be interpreted, but I’m interested in keeping an eye on this,” tweeted Calli Schroeder, Global Privacy Counsel at the Electronic Privacy Information Center (EPIC).

Italy bans facial recognition unless it is used to fight crime. This would affect all private use.

Obviously some big questions about how broadly “crime fighting” will be interpreted, but interested to check this out.https://t.co/fEVc3nxmEY

— Techni-Calli (@Iwillleavenow) November 15, 2022

Italy’s new rules come amid heated debates in the EU over FRT restrictions. The block is moving towards milestone approval AI lawwhich would become the world’s first comprehensive legal framework for artificial intelligence.

FRT has proven to be a divisive component of the proposals. A majority of EU legislators now want to ban using the random and real-time crowd scanning technology. However, these restrictions are set so that there are exceptions for the police.

Supporters of these dispensations claim so help fight crimebut opponents warn against this they are ripe for abuse. In September, Dragos Tudorache, Member of the European Parliament (MEP) and co-rapporteur on the AI ​​law, said the exemptions would create “very difficult control and accountability”.

We would – and I will do my best – get a full ban in Parliament.

Critics are wary of the rules that lead to mass surveillance and oppressive police action. These risks are particularly acute for marginalized groups and dissidents, who are disproportionately victims of discriminating AI and inaccurate FRT results.

These results are attributed to biased policing, unrepresentative training data, and a combination of the two. Campaigners fear that FRT is algorithmic amplify and automate human prejudices.

These dangers have also raised the concerns of lawmakers. Last week, Brando Benifei, an Italian MEP and co-rapporteur on the AI ​​law, called for the EU’s new regulation to include an absolute ban about biometric mass surveillance.

“Today there are two loopholes in the ‘ban’ proposed by the Commission: private spaces and ‘ex-post’ recognition are not covered, but there are also exceptions related to some criminal investigations and prosecutions,” he said. “We would – and I will do my best – have a full ban in parliament.”

The final rules are still working through the EU, but agreement could be reached by mid-November. As the world’s first attempt to horizontally regulate AI, the impact will extend beyond the bloc.