Canadian privacy & access decisions

The Office of the Privacy Commissioner of Canada (OPC) initiated complaints against X Corp. and X.AI LLC following reports that their AI chatbot, Grok, generated millions of sexualized deepfakes of identifiable individuals. The investigation focused on whether valid consent was obtained for the collection, use, and disclosure of personal information for this purpose, and if such practices were appropriate under PIPEDA. The OPC found that neither company obtained valid consent, noting the sensitive nature of the information, the unreasonableness of individuals' expectations, and the significant risk of harm. Furthermore, the OPC concluded that the generation of sexualized deepfakes was inappropriate, as the loss of privacy and harm far outweighed any benefits, and less privacy-invasive means were available. While the companies implemented some safeguards, the OPC deemed their initial response insufficient and their current measures unproven. Consequently, the matter was found well-founded, with the OPC making several recommendations for improved safeguards, proactive monitoring, and annual third-party audits, while committing to ongoing monitoring.

• Whether PIPEDA applies to X Corp. and X.AI LLC, specifically regarding the existence of a "real and substantial connection" to Canada.
• Whether deepfakes of identifiable individuals, including sexualized deepfakes, constitute "personal information" under PIPEDA.
• Whether X Corp. and X.AI LLC obtained valid consent for the collection, use, and disclosure of personal information to generate sexualized deepfakes, as required by Principle 4.3 of PIPEDA.
• Whether express consent was required for the generation of sexualized deepfakes, considering the sensitivity of the information, individuals' reasonable expectations, and the risk of significant harm (Principle 4.3.4, 4.3.5, and s.6.1 of PIPEDA).
• Whether X Corp. and X.AI LLC are accountable for ensuring valid consent for content generated by their tools in the course of commercial activity.
• Whether a reasonable person would consider the collection, use, and disclosure of personal information for the purpose of an image generation service capable of producing sexualized deepfakes to be appropriate in the circumstances, as per subsection 5(3) of PIPEDA.
• Whether the organizations had a legitimate need or bona fide business interest that extended to providing an image generation tool capable of producing non-consensual sexualized deepfakes.
• Whether less privacy-invasive means were available to achieve the organizations' purposes at comparable cost and benefits.
• Whether the loss of privacy and risk of harm associated with sexualized deepfakes were proportionate to the benefits of the practice.
• Whether X Corp. and X.AI LLC's initial response and implemented safeguards were sufficient and effective in preventing the generation of sexualized deepfakes.