Law enforcement in Canada is increasingly turning to facial recognition in hopes of augmenting their investigative powers. Facial recognition is the process of identifying a person or verifying their identity on the basis of facial data and patterns. There are numerous accuracy challenges associated with facial recognition technology that can exacerbate historical prejudices and stereotypes, especially when deployed at large scale. Studies demonstrate that facial recognition algorithms discriminate against elderly people, children, women, racialized people, as well as the LGBTQ2S+ community. Overconfidence in the technology can lead to serious, and at times devastating, consequences for marginalized individuals.
The technology also threatens the right to anonymity, privacy, and substantive equality. The police can use facial recognition to arrest someone after an alleged crime has occurred by comparing images with a watchlist or general image database. They can also conduct the same comparisons in a live setting for example through CCTV cameras, tracking people’s locations and movement in real time without the knowledge or consent of those being surveilled. In either case, facial recognition poses significant threats to privacy, fundamental freedoms, and other human rights.
It is in this context that the RCMP decided to use the services of data scraping and facial recognition company Clearview AI on a trial basis, ultimately leading Canada’s federal, provincial and territorial privacy protection authorities to jointly develop guidance for police agencies across Canada on facial recognition. The guidance document outlines the current state of the law in Canada regarding police use of facial recognition and encourages best practices around privacy impact assessments, accuracy, data minimization, purpose limitation, data security, retention, openness and transparency, individual access, and accountability.
We welcome the opportunity to provide feedback on this guidance document. The only commentary we provide on the guidance itself is the encouragement to consider recommending that law enforcement use decentralized, rather than centralized, databases in order to reduce the risk of the systems being compromised or repurposed as explained by privacy and human rights expert Tamir Israel — whose work examining the legal aspects of facial recognition is foundational in the Canadian context. The remainder of our feedback takes the form of comparative analysis of the treatment of facial recognition in other jurisdictions and recommendations for advancing Canada’s privacy legal framework to address law enforcement’s use of facial recognition technology.
Given the substantial risks posed by this technology, facial recognition systems should not be adopted at this time and the proportionality of current systems in use by law enforcement should be reassessed and reexamined. However, adoption of any facial recognition system for identification requires a dedicated legislative framework that prohibits the use of the technology in the absence of explicit lawful authority and that enables the Office of the Privacy Commissioner of Canada (OPC) to provide oversight of these systems through robust requirements, such as ongoing privacy and algorithmic impact assessments that are made available to the public.
Author: Yuan Stevens | October 2021