The Global Network Initiative (GNI), a global multistakeholder organization focused on freedom of expression and privacy in the technology sector, is concerned about certain amendments recently introduced in the proposed Artificial Intelligence and Data Act (AIDA) in Canada, which is adjoined to the Bill C-27, The Digital Charter Implementation Act. The government of Canada, through the Minister of Innovation, Science, and Industry, introduced a series of amendments to AIDA at the end of last year. Despite the fact that these amendments represent significant changes, they were introduced without prior consultation. GNI recommends additional consultation on AIDA, and in particular regarding two recently-proposed provisions that raise freedom of expression and privacy concerns.
The aforementioned proposed amendments include a list of uses that are considered “high-impact,” including the use of an artificial intelligence (AI) system in content moderation, including by search and social media services. Digital services of all sorts use AI systems in a wide variety of ways to manage content. Categorizing such a broad use case as high-impact raises significant questions about how “content moderation” might be defined and applied in practice, and whether a narrower approach would be more appropriate. GNI has cautioned against risks related to automated approaches to content moderation and supports transparency and risk assessment-based approaches to technology governance generally. However, categorizing all content moderation uses of AI as high-risk, and therefore requiring that regulatory requirements be met prior to their introduction, AI tools may unnecessarily limit the ability of services to proactively address various types of harms. GNI notes that while several of the introduced amendments function to bring AIDA in line with international definitions and approaches to AI regulation, this category is not identified as high risk in the EU AI Act. Furthermore, the subsequent introduction by the government of the “Online Harms Act,” which would regulate many of the same services, raises significant questions about the extent to which that bill, including its proposed Digital Safety Commission, Ombudsman, and Office, might overlap with AIDA creating the potential for confusion, inconsistency, and duplication.
These concerns are exacerbated by the significant powers that the amendments would provide to the newly-proposed AI and Data Commissioner. These include authorization to “enter any place other than a dwelling-house and examine anything found in the place, use or cause to be used any computer system at the place, including for the purpose of gaining access to any electronic data,” including through “remote access.” Remote government access to data is often associated with intrusions on privacy and other surveillance-related concerns. Instead, GNI recommends that the government consider approaches that allow for data and systems to be “sandboxed” for review, with sufficient oversight and accountability provisions to ensure such access is not abused.
Canada is a global leader on matters of innovation, human rights, and AI. The approach that Canada takes on tech regulation will serve as a model and inspiration for many countries around the world. In order to ensure that its regulatory approach continues to foster and uphold that reputation and set a positive example globally, GNI recommends that the government and Parliament seek further opportunities to consult and engage a wide-range of experts, including from academia, civil society, and industry on AIDA generally and these provisions in particular. GNI stands ready to continue engaging in such consultations.