In Doan v. Clearview AI Inc., 2025 FCA 133, released on July 16, Justice Elizabeth Walker rejected arguments from Clearview that a class definition that effectively requires a class member to take steps to obtain information regarding their status is unacceptable.
“The fact there will almost inevitably be Class Members who do not engage with the process does not result in an opt-in process, nor does it thwart the objective of the class action regime in promoting access to justice. The Class Member remains part of the class and could, if they so wish, take steps to opt out. They remain unengaged at their own risk,” the judge wrote.
The proposed class action was brought by appellant Ha Vi Doan, who alleged that the respondent Clearview AI Inc. infringed copyright in photographs she had taken by using web crawlers to download, store and index images posted online.
Clearview AI uses photographs downloaded online to provide facial recognition services to its clients, predominantly law enforcement and national security agencies.
The company is based in the United States and operated in Canada between 2019 and July 2020. It ceased offering its services in Canada after federal and provincial privacy protection authorities initiated investigations into its operations.
The appellant has alleged that Clearview engaged in widespread copyright infringement and violations of moral rights, by collecting, copying and storing images of human faces taken in Canada without the consent of the individuals who took the images or who hold copyright and/or moral rights in those images.
The proposed class was defined by the appellant as including all Canadian residents or citizens who are authors of photographs collected by Clearview AI and who have not assigned or licensed their copyrights, as well as individuals or entities to whom those copyrights had been assigned or licensed.
In Doan v. Clearview Inc., 2023 FC 1612, the Federal Court refused to certify the proposed class action, finding that the appellant had failed to meet the certification requirement of an identifiable class of two or more persons.
The certification judge found that class members could not self-identify on the basis of the class definition and that the defendant could not identify class members based on metadata in its database.
The appellant had proposed that class members could submit queries to Clearview to confirm their class membership.
The judge held that this methodology would transform the opt-out class action scheme into an opt-in scheme and refused to certify the action as a class proceeding.
The certification judge also observed that Clearview does not respond to queries from Canada or queries that require a search for third parties.
Doan appealed the decision, arguing that the certification judge erred in concluding that the proposed class member identification methodology would transform the opt-out class action scheme into an opt-in scheme.
Justice Walker noted that the Federal Court’s class proceeding rules clearly establish an opt-out scheme that permit an individual who does not wish to be part of a class proceeding to opt out of the proceeding and pursue their own recourse.
The court observed that case law cited by the certification judge supported the importance of the right to opt out but did not suggest that a query mechanism necessarily converts a proposed class proceeding into an opt-in scheme or otherwise undermines the opt-out nature of the proceeding.
The judge noted that the pleadings in the proposed class action contemplated an opt-out class action scheme.
“If a person does nothing, they remain a Class Member assuming they later prove they are a Class Member. Similarly, an individual who is unsure whether they are or are not a Class Member and who submits a query to Clearview remains a Class Member unless they choose to opt out, again assuming they later prove they are a Class Member,” she wrote.
The court found that the fact that class members choose to confirm that status via a query to Clearview did not change the characterization of the proceeding as an opt-out scheme.
The court also rejected Clearview’s submissions that a class definition that effectively requires a class member to take steps to obtain information regarding their status is unacceptable.
The appellant had also argued that the certification judge erred in accepting Clearview’s statement that copyright information in the photographs it collected was not retrievable by its web crawlers.
Doan had also submitted at the certification hearing that class members could, under the U.S. Digital Millennium Copyright Act, upload their images, compare them to images in the database and request that Clearview remove their images from its database.
Doan argued that the certification judge misapprehended or overlooked evidence regarding two image identification methods Clearview uses, one of which was available to Canadians in 2020 and 2021.
Justice Walker observed that Clearview had, in its written submissions, argued that neither query method could identify class members during the opt-out period but could be used to prove identity later in the proceedings, at the discovery or recovery stages.
The court observed that it was not clear why this process could only be used later in the process to prove identity and that Clearview’s counsel had on being questioned conceded that class members could in theory query Clearview to obtain sufficient information during the opt-out period.
Justice Walker held that the fact that a query method may be available to identify class members during the opt-out period must be fully canvassed.
She concluded that the certification judge erred in concluding that the proposed query method to identify class members transforms the underlying action into an opt-in scheme.
The court further found that the certification judge did not fully examine whether the evidence demonstrated some basis in fact of a workable method to identify class members before the end of the opt-out period.
The court allowed the appeal and returned the certification motion to the Federal Court for redetermination.
Justices Richard Boivin and George Locke concurred in the decision.
Counsel for the parties were not immediately available for comment.
Counsel for the appellant were William Colish and Mélina Cardinal-Bradette of Alexeev Attorneys Inc. and Michael Shortt of Fasken Martineau Dumoulin LLP.
Counsel for the respondent were Olga Redko, Mathilde Couture and Marie-Hélène Lyonnais of IMK LLP.
If you have any information, story ideas, or news tips for Law360 Canada on business-related law and litigation, including class actions, please contact Karunjit Singh at karunjit.singh@lexisnexis.ca or 905-415-5859.