Hungary on EU watchlist over surveillance at Pride

BRUSSELS — The European Union is looking into whether Hungary’s plan to deploy facial recognition technology to identify people attending LGBTQ+ Pride events is illegal.

The move by the EU, reported exclusively by POLITICO, sets up the latest clash between Brussels and Prime Minister Viktor Orbán’s government as Hungary becomes an increasingly troublesome member of the bloc.

It also marks the first real test of landmark EU artificial intelligence rules that critics say contain too many exemptions to effectively protect against abusive uses of technology. 

Hungary in mid-March banned Pride gatherings and changed its laws to allow police to use biometric cameras to identify protestors who attend such events. The government said it had done so to protect children from the LGBTQ+ agenda, but opponents say Budapest is stoking an anti-LGBTQ+ campaign to garner far-right support — and is hitting the limits of EU legality.

The European Commission is “currently assessing” the Hungarian law to see whether it complies with the bloc’s AI Act, Commission spokesperson Thomas Regnier told POLITICO.

One of the things the Commission needs to decide is whether facial recognition is happening in real time — which would constitute a breach of newly implemented AI rules — or whether the ways in which surveillance images are being processed gives Hungary an opt-out.

The AI Act foresees administrative fines of up to €35 million for violations, but it’s up to EU countries themselves to appoint an authority to impose a fine. Still, EU countries have to report back to the Commission if they allow some of the prohibited practices, which gives Brussels some power to intervene.

Regnier said the Commission “will not hesitate to take action, where appropriate.”

A Hungarian government spokesperson said Hungary “believes all is in line with our constitution and EU law.”

Growing pressure

The EU’s executive has been under heavy pressure to respond to Hungary’s Pride ban, particularly on the use of facial recognition.

The EU moved in February to prohibit police forces from using real-time facial recognition technology under its artificial intelligence rulebook. Whether to ban systems that can identify people in real time based on their biometric data (i.e. through CCTV) was one of the thorniest topics as the law was being agreed — European Parliament lawmakers largely favored a ban while EU countries wanted an opt-out to allow them to fight serious crime.

As a result, “real-time biometric identification” by police forces is forbidden, but with significant exceptions. 

“Hungary’s amended law appears to enable the use of real-time biometric surveillance at Pride events, which is an application clearly falling under the AI Act’s prohibitions,” said Italian social democratic European lawmaker Brando Benifei, one of the two architects of the EU’s AI Act.

“Such systems are not only technically designed to identify individuals in public without consent, but also risk inferring sensitive attributes like sexual orientation,” he said in a statement shared with POLITICO. 

Two dozen civil society groups expressed their concerns about Hungary’s move in a letter to Commission President Ursula von der Leyen last week.

On April 14, the Commission's Executive Vice President for Tech Sovereignty Henna Virkkunen met with the new Hungarian government commissioner for AI, László Palkovics.

Commission spokespeople didn't say whether Virkkunen had raised the issue with Palkovics.

Palkovics served as an independent tech and innovation minister in previous Orbán governments. He was appointed in late February and will be in charge of setting up the Hungarian authority that will oversee enforcement of the AI Act. 

A matter of seconds

Hungary can avoid an AI Act ban if facial-recognition technology is not used in real time but has a time lag — which, although not forbidden, is considered a “high-risk” application.

That could well be the case, said Ádám Remport, a legal expert at the Hungarian Civil Liberties Union, which campaigns in favor of human rights and opposes the government’s plan. 

According to Remport, the Hungarian police take footage from public cameras and send it to the Hungarian Institute for Forensic Sciences, which does the actual facial recognition.

“Because there's a time lag here, it may not be real time according to the AI Act, but we don't know what that time lag is,” he said. 

Even if there’s a time lag, Hungary will still have to adhere to “strict regulatory requirements,” said the Commission’s Regnier, such as conducting an impact assessment.

That would also require the Hungarian government to seek approval from a judge — but Brussels and Budapest are engaged in an ongoing tug-of-war over the independence of judges and the country’s adherence to the rule of law.

For campaigners, it’s proof that the EU’s AI regulation has too many loopholes that allow police forces to use facial-recognition technology. 

“We are now seeing the concrete consequences of what happens when we succeed in watering down certain regulations,” Remport said.