European Parliament strengthens protection of citizens’ fundamental rights in AI Act vote

11 May 2023

Draft legislation recognises the significant risks posed by facial recognition technology

ICCL has welcomed the outcome of a vote in the European Parliament this morning on the EU’s draft Artificial Intelligence (AI) Act. The vote is a significant step towards finalising the Parliament’s position on the Act and represents a blow to the Irish Government’s plans to introduce facial recognition technology (FRT) for An Garda Síochána.  

Speaking today, Dr Kris Shrishak, Senior Fellow, ICCL said:  

“The EU’s AI Act could become a global standard for regulation of AI systems. The stakes simply couldn’t be higher for people’s fundamental rights. Today’s vote is the result of extensive negotiations and compromises and sends a clear message that the Parliament feels AI should be strictly regulated to protect EU citizens’ rights. 

"The Parliament has fixed many loopholes and has brought AI systems like ChatGPT within scope. But there are still many concerns. The Parliament has failed to take a firm stand on preventing companies from gobbling up data for free, including copyright-protected data. It only requires companies to provide a summary of copyright-protected data used for training AI systems. 

“The Parliament has also tied the regulators’ hands by preventing them from accessing the source code of AI systems for investigations. We are concerned that we will see a repeat of GDPR-like enforcement problems.”  

 

The AI Act, Facial Recognition Technology (FRT) and Policing 

In today’s vote, two committees of the European Parliament agreed a text which would ban: 

  • All use of live FRT in public spaces, with no exceptions; 
  • Running an FRT analysis of recorded footage obtained from public spaces, with the exception of law enforcement for the prosecution of serious crimes and only with judicial approval in a specific case. 

Olga Cronin, Senior Policy Officer, Surveillance and Human Rights, ICCL, said:  

“Today’s vote in the European Parliament has very serious implications for the Government’s plans to introduce facial recognition technology for Gardaí. The Parliament has recognised the significant risks inherent in FRT and definitively moved to protect the rights of EU citizens.  

“The decision to ban live FRT in public spaces is very welcome, but we remain concerned that in certain circumstances law enforcement would be permitted to use retrospective FRT under this draft legislation. Retrospective use of FRT represents as major an interference with people's fundamental rights as live FRT use – and it has been found to be potentially even more risky for people’s rights. 

"FRT systems risk rights because they can allow for disproportionate mass surveillance and turn us into human licence plates by enabling the identification and tracking of individuals without warranted suspicion. They can have a chilling effect on people’s behaviour and have been shown to be biased and discriminatory, particularly for anyone who is not a white man. In addition, ICCL is not confident that full data protection measures for the collection and processing of biometric data are in place at An Garda Síochána.” 

Ends

Available for comment:

  • Dr Kris Shrishak, Senior Fellow, ICCL (for queries related to the AI Act)
  • Olga Cronin, Senior Policy Officer, Surveillance and Human Rights, ICCL (for queries related to facial recognition technology (FRT) in Ireland)

For media queries: ruth.mccourt@iccl.ie / 087 415 7162

Notes to Editors

ICCL made a significant number of recommendations on how to improve the AI Act. Read details of our recommendations and those which were included in the Act here.

The European Parliamentary Research Service has previously warned that the risk of persistent tracking, and its associated adverse impact on fundamental rights and democracy in respect of retrospective use of remote biometric identification systems, such as FRT, is “at least equivalent” to the risk associated with 'real-time' remote biometric identification. Noting the weakness of the distinction between the two, it warned (page 55):

As the images potentially available for 'post' remote biometric identification of natural persons are actually more numerous than those available at any point in time for real-time identification, they should also make it possible to draw a much more complete picture of the activities of any individual, thus representing a major interference with their fundamental rights.

Read the full study here.