Laws and Loopholes

The European Union this week moved to ban some “unacceptable” uses of Artificial Intelligence (AI), setting the world’s first limits on the technology, even as critics said the bloc didn’t go far enough, Politico reported.

With the AI Act, the bloc became the first national or supranational entity to attempt to regulate the technology, banning AI’s use to profile future criminals with “predictive policing.” It was one of seven categories the bloc banned as of Feb. 2, including the practice of scraping images from the Internet to build a database for facial recognition.

The AI Act aims to ensure the “trustworthy” use of AI in Europe, the European Commission explained.

The rules set the EU apart from other countries or regions where there are few restrictions on the technology, underscoring its usual role as safeguarding the public from abuse of technology. The bloc has long been at the forefront of Internet privacy rights.

The bloc will create an enforcement authority for AI rules by August. Officials also said they would further define and enhance the rules over the next 18 months.

Italian lawmaker Brando Benifei, who was involved in the legislation, said that the bans are linked “to the protection of our democracies.”

There have been cases of AI run amok that led the EU to consider such restrictions.

For example, in 2019, Dutch tax authorities used an algorithm to spot suspected benefit fraud and ended up wrongly accusing about 26,000 people of fraud. That practice – predictive policing – is banned under the new rules.

Digital rights campaigners said the law still has too many loopholes, saying that police and migration authorities – who are technically not allowed to use certain tools – will be able to secure exemptions from EU countries to continue using AI features such as real-time facial recognition in public places, especially when it comes to the most serious crimes.