Compromise AI Act Would Drop Biometric Ban

The European Parliament is reportedly close to reaching a compromise regarding the use of ‘remote biometric identification’ (RBI) technologies, as part of the negotiations for the new AI Act. The draft law, now in the final stages of the EU legislative process known as trilogues, has been a subject of contention, particularly around the use of RBI in real-time.

On November 3, a compromise text circulated by the offices of European Parliament’s co-rapporteurs, Dragoș Tudorache and Brando Benifei, indicated a shift from a complete ban on real-time RBI to allowing it under strict conditions. This followed a similar proposal by the Spanish presidency of the EU Council of Ministers. The compromises suggest that RBI could be used for locating serious criminal suspects accused of crimes punishable by a minimum of five years, including terrorism and other severe offences.

The proposed adjustments require law enforcement to register RBI systems in an EU public database and complete a fundamental rights impact assessment prior to use. Real-time RBI operations would typically need validation by a judicial authority, although in exceptional cases, authorization could be requested retrospectively.

In exchange for these concessions, the EU Parliament has secured an extension of the list of prohibited AI applications. Inspired by the Clearview AI case, there will be a ban on AI systems that scrape facial images on a large scale and on those using biometric categorization to infer sensitive personal data. Additionally, emotion recognition technology would be forbidden in work and educational contexts.

The EU Council has proposed significant exemptions for law enforcement agencies, such as omitting the requirement to report serious incidents identified in high-risk AI systems, and allowing the use of high-risk AI applications without a prior conformity assessment in exceptional circumstances. Provisions to protect sensitive operational data from disclosure have also been included.

Lastly, an exemption related to national security has been a focus, with France advocating for an extensive exclusion of military, defense, or national security systems from the AI Act’s scope. The EU Commission had expressed concerns about this exemption, but compromise wording has been included in the latest draft, potentially aligning it with the national security exemption of the Data Act.

News of the compromise draft comes alongside an open letter from Volker Türk, the United Nations High Commissioner for Human Rights, urging the European Union to ensure that its AI Act is firmly grounded in human rights law. The High Commissioner advocates for strict limits on biometric surveillance and crime prediction tools, endorsing the European Parliament’s call for a ban on biometric tools that categorize individuals by protected characteristics. Furthermore, Türk supports comprehensive fundamental rights impact assessments to ensure AI practices respect human rights and offers the UN’s assistance in perfecting the AI Act to safeguard individual rights.

Source: Euractiv

–

November 10, 2023 – by the FindBiometrics Editorial Team