Know better. Do better.

With artificial intelligence (AI) developing at breakneck speed, policy makers have struggled to keep up. But the European Union (EU) was quick out of the traps in August when the world’s first comprehensive set of regulations on the use of the technology entered into force.

The new rules categorise AI systems by their capacity to cause harm to society - the higher the risk, the stricter the rules, with some systems set to be banned entirely within the EU by February 2025.

Companies that do not comply face hefty fines.

In the words of EU lawmaker Brando Benifei, who helped shaped the AI Act: “While Big Tech companies are sounding the alarm over their own creations, Europe has gone ahead and proposed a concrete response to the risks AI is starting to pose.”

But, in the rush to install guardrails and in the midst of sticky negotiations with governments and Big Tech, exemptions were made for some of the most controversial uses of AI by police and border authorities, including facial and emotion recognition tech.

Given that, activists say, for people arriving at Europe’s increasingly digitally-fortified borders, AI could become yet another brick in the wall.

As political rhetoric around immigration hardens across the continent, Europe is pouring billions into border management, increasingly turning to new tech to boost security, from AI-driven watchtowers and lie detector tests to algorithms that can predict migration flows.

But, with the AI Act now in play, allowing exceptions at Europe’s borders could boost unlawful surveillance, discrimination, and criminalise people on the move, say rights campaigners.

“Instead of crafting solutions to protect vulnerable people … policies are being made that foster suspicion,” said Caterina Rodelli, a policy analyst with digital rights group Access Now, who recently visited Greece, which has become one of the main testing grounds for the new tech.  

“There’s a tendency in Europe to rely on technology to go as fast as possible,” Romain Lanneau, a legal researcher from human rights charity Statewatch, told me. 

But while AI tools may boost efficiency as the EU pushes to fast-track and deport unsuccessful applicants, activists say it also represents a risk for vulnerable people. 

Asylum seekers may end up rejected by algorithms that could replicate or amplify existing biases, all while giving a veil of machine-made objectivity, Rodelli said. 

Campaigners are also raising the alarm over the use of algorithmic tools that can forecast migration flows and may help facilitate people being violently and illegally pushed back from borders. 

An MSF report in 2023 described how asylum seekers in Greece faced: "physical violence, including being beaten, handcuffed, strip-searched, having their possessions confiscated, and forcibly sent back to sea”. 

Greece’s government has denied it engages in pushbacks.

Another criticism is that the new rules don’t prevent EU-based companies developing harmful AI systems for export with no thought of how that technology might contribute to abuses elsewhere.

"If Europe wants to be a standard-setter globally on a regulation that is human rights-centred, I don't think it's sending the right message," Mher Hakobyan, Amnesty Tech's advisor on AI regulation, told me.

Though rights groups see the AI Act as a missed opportunity to protect all people against the harms of AI, they say their battle is not over. 

Expect lawsuits, says Lanneau.

In April, the Greek government was slapped with a fine over AI surveillance in refugee camps when it was found to have violated data privacy regulations. And, in 2021, the Netherlands was gripped by a scandal in which the use of an algorithm led to parents, most of whom were from ethnic minority backgrounds, being wrongly accused of child benefit fraud. 

It may be, campaigners say, that legal challenges and public backlash are what will ultimately close the human rights gaps.