Authors: Chris Tenove, Spencer McKay, and Heidi Tworek
Generative artificial intelligence (AI) is widely seen as a potential threat to election integrity and democratic participation. AI-enabled content creation and dissemination have not yet decisively changed election outcomes, but experimental evidence and real-world cases suggest that AI is likely to be a major challenge for democracy in Canada and globally.
This article uses a multiple-streams framework to analyze potential policy responses in Canada. First, we clarify the problem of generative AI as a threat to democratic elections by specifying three harmful uses: deception, information system pollution, and harassment. Second, we examine potential responses in the policy stream in Canada and globally, which we categorize as disclosure measures, content prohibitions and takedowns, and comprehensive regulations for AI systems. We argue that a governance ecosystem, rather than a single policy, must be created to achieve substantial accountability for AI-enabled harms. Third, our analysis of the politics stream suggests that some disclosure and content-focused policies are politically feasible in the near term but that more comprehensive regulation is unlikely without a significant change in the political context.
Read the full, open-access report here.