AI Content Moderation Systems Reported to Disproportionately Remove French, Indigenous, and Racialized Content

AI-powered content moderation systems deployed by major social media platforms operating in Canada have repeatedly demonstrated disproportionate error rates when processing content in French, Indigenous languages, and content from racialized communities. According to whistleblower Frances Haugen's 2021 congressional testimony, internal documents from Meta indicated that approximately 87% of the company's global misinformation spending was allocated to English-language content, even though English speakers represent roughly 9% of the platform's user base (Rest of World, 2021). Haugen characterized this as an approximate figure. This figure reflects Meta's global resource allocation and has not been independently verified for Canadian operations specifically. Non-English languages — including French — received substantially less investment in classifier training and human review capacity (Rest of World, 2021). This pattern extends across platforms: automated systems trained predominantly on English-language data frequently misclassify content in other languages, leading to both over-removal of legitimate speech and under-removal of harmful content (CBC News, 2021; Citizen Lab, University of Toronto, 2021).

Francophone Canadians — particularly in Quebec — use social media platforms where moderation systems may misinterpret Quebecois vernacular, colloquialisms, and cultural context. Indigenous language speakers face even starker gaps: content in Inuktitut, Cree, Anishinaabemowin, and other Indigenous languages likely receives minimal moderation coverage, given that these low-resource languages have little or no representation in platform training data. The House of Commons Standing Committee on Canadian Heritage, in its November 2024 report on "Tech Giants' Intimidation and Subversion Tactics to Evade Regulation," examined how major platforms resisted Canadian regulatory efforts, including through news access restrictions and lobbying campaigns (House of Commons Standing Committee on Canadian Heritage, 2024).

The pattern is ongoing rather than a single event. The Citizen Lab at the University of Toronto, in its submission on the federal government's proposed approach to online harms, noted that people in Canada access content in hundreds of languages and dialects that do not receive equal moderation resources from platforms (Citizen Lab, University of Toronto, 2021). Haugen's testimony and subsequent reporting suggested that platforms invest moderation resources roughly in proportion to advertising revenue rather than user population or rights impact, meaning languages and communities with less commercial value may receive worse service (Rest of World, 2021; CBC News, 2021). In the Canadian context, commentators have raised questions about how the Official Languages Act's guarantee of linguistic equality applies to digital platforms where an increasing share of civic discourse occurs.