EU pols divided on AI Act as deadline nears
As the European Parliament prepares to vote on the landmark Artificial Intelligence Act this month, its members are still negotiating bans and high-risk judgments for predictive policing, emotion recognition and biometrics in public spaces.
Lawmakers are already expecting pushback from European Union member states and their local police forces over a proposal to ban certain AI uses from security and policing roles, including facial recognition in public spaces.
But criticism may come from more than member states. Some members of parliament (MEPs) are not keen on all of the proposed bans, according to EU Observer.
Among the restive factions are the center-conservative European People’s Party and liberal lawmakers from Renew Europe. The latter party includes Romania MEP Dragoș Tudorache, who, along with Italy socialist Brando Benifei, leads the committee conducting AI Act passage negotiations.
Bans are viewed more favorably by the Greens and socialists.
Among the topics being negotiated by Tudorache and Benifei’s committee involves supervision policies governing police agencies that use biometric recognition.
The presumed final Parliament vote on the AI Act draft will take place May 11. Tudorache reportedly told the EU Observer last week that he would be holding his breath “until the last second.”
If the vote is successful and the Parliament adopts it, the Council of the European Union, which represents member states, and parliamentarians are expected to start negotiating the final text of the legislation in a trialogue with the European Commission.
The Council adopted a common position on issues including biometrics in public last December. It has stopped short of an outright ban on AI for policing and security.
Group wants AI Act tech standards panels to add human rights experts
Civil society groups have generally welcomed the parliament’s first draft of the act but others with human rights organization Article 19 have called for more attention to another part of AI legislation — the committees setting technical standards for high-risk AI.
Article 19 includes organizations such as European Digital Rights, Access Now, Algorithm Watch, Amnesty International and others. Last week, the group called for a ban on AI systems that pose an unacceptable risk for fundamental rights, including all types of remote biometric identification.
Those groups are advocating for the inclusion of human rights and civil society experts in the committees working on technical frameworks, requirements and specifications for key high-risk technologies.
Authors of the AI Act have designated high-risk AI missions as those used in education, employment, public benefits programs and law enforcement. The use of biometrics in public spaces is another. Any developer that wants to bring high-risk AI into the EU would need to demonstrate compliance with the act’s “essential requirements.”
Two technical standards organizations, European Committee for Standardisation (CEN) and the European Committee for Electrotechnical Standardisation (CENELEC) will determine these rules, which likely will address data governance, transparency, security and human oversight.
“These bodies are almost exclusively composed of engineers or technologists that represent EU member states,” Article 19 representatives Mehwish Ansari and Vidushi Mardais write in an opinion piece for the EU Observer.
“There is a real danger that these bodies will have the de facto power to determine how the AI Act is implemented without the means to ensure that its intended objective — to protect people’s fundamental rights — is truly met,” they wrote.
The Ada Lovelace Institute in April voiced Similar warnings.
AI | AI Act | biometric identification | EU | facial recognition | legislation