‘Urgent clarity’ sought over racial bias in UK police facial recognition technology | Facial recognition
The UK’s data protection watchdog has demanded “urgent clarity” from the Home Office on racial bias in police facial recognition technology before considering next steps.
The Home Office admitted the technology was “more likely to inadvertently include some demographic groups in search results” following testing of its application to the police’s national database by the National Physical Laboratory (NPL).
The report found that technology intended to be used to catch serious criminals was more likely to incorrectly match black and Asian people than their white counterparts.
In a statement responding to the report, Information Commissioner’s Office deputy commissioner Emily Keaney said the ICO was seeking “urgent clarity on this matter” from the Home Office for the watchdog to “assess the situation and consider our next steps”.
Next steps could include enforcement actions, including issuing a legally binding order to halt the use of the technology or fines, as well as working with the Home Office and police to make improvements.
Keaney said: “Last week, we became aware of historical bias in the algorithm used by forces across the UK for retrospective facial recognition in the police’s national database.
“We accept that measures have been taken to address this bias. However, it is disappointing that we were not made aware of this sooner, despite meeting regularly with the Home Office and police agencies as part of our wider work to hold government and the public sector to account for how data is used across their services.”
“While we appreciate the valuable role technology can play, public confidence in its use is crucial and any perception of bias and discrimination can increase distrust. The ICO is here to support and help the public sector get this right.”
Police and crime commissioners said the publication of the NPL’s finding “shed light on a relevant bias” and urged caution against national expansion plans that could include installing cameras in shopping centres, stadiums and transport hubs without adequate security measures.
The findings were announced on Thursday, hours after policing minister Sarah Jones described the technology as the “biggest breakthrough since DNA matching”.
Facial recognition technology scans people’s faces and compares the images to watch lists of known or wanted criminals. It can be used to review live footage of people passing by cameras, compare their faces with those on a wanted list, or enable police officers to target people passing by mounted cameras.
Police officers may also retrospectively review images of suspects in police, passport or immigration databases to identify suspects and check their backgrounds.
Examining the police national database’s retroactive facial recognition technology tool in a lower setting, analysts found that “the false positive identification rate (FPIR) for white subjects (0.04%) was lower than for Asian subjects (4.0%) and black subjects (5.5%).”
The testing found that the number of false positives was particularly high for black women. “The FPIR of black male subjects (0.4%) was lower than that of black female subjects (9.9%),” the report said.
Responding to the report, a Home Office spokesperson said the department was taking the findings “seriously” and had already taken action, including procuring and testing a new algorithm that “does not have statistically significant bias”.
“Given the importance of this issue, we have asked the police inspectorate, as well as the forensics regulator, to review law enforcement’s use of facial recognition. They will assess the effectiveness of mitigating measures supported by the National Police Chiefs’ Council,” the spokesman said.
