Contenu généré par l'IA et réseaux de robots ayant ciblé l'élection fédérale canadienne de 2025

Canada's 2025 federal election saw AI-generated content and automated amplification operating at documented scale across multiple simultaneous vectors — a qualitative shift from previous Canadian elections.

The Atlantic Council's DFRLab identified highly active bot-like accounts on X that amplified political content in a spam-like manner ahead of the April 28 election, frequently replying to posts from federal parties and their leaders (DFRLab, 2025). A Financial Times investigation separately identified a coordinated network of suspicious accounts favouring Poilievre and attacking Carney. DFRLab's analysis found that approximately 80% of the politically charged spam and misleading narratives from bot-like accounts were directed at the Liberal Party and its leadership (DFRLab, 2025). The pattern was consistent with coordinated inauthentic behaviour that could distort perceived political sentiment.

AI-generated fabricated images were created and circulated to manufacture false political associations. These included an AI-generated image depicting Carney with Jeffrey Epstein in a pool, which appeared on X on January 27, 2025 and was debunked by fact-checkers the following day (CTV News, 2025). A separate AI-generated image depicting Carney dining with Ghislaine Maxwell was documented after the election (earliest appearance May 3, 2025) (CTV News, 2025). These fabrications were designed to seed conspiracy narratives about Carney's associations. A deepfake video manipulated authentic footage of a March 27 press conference to falsely show Carney announcing a ban on vehicles made before 2000, reaching millions of views on TikTok and X (CTV News, 2025). Separately, deepfake videos mimicking CBC news interviews were used to direct viewers to cryptocurrency scam websites — financial fraud documented in the related Carney deepfake record. These fabrications spread through both bot amplification and organic sharing, with conspiracy narratives gaining traction across multiple platforms (DFRLab, 2025).

A website called "Pierre Poilievre News" published AI-generated articles filled with unverified information presented as legitimate political journalism (CTV News, 2025). At the end of March 2025, a fabricated claim from this site — asserting that Poilievre's personal fortune exceeded $25 million — spread widely on social media (CTV News, 2025). The site produced content designed to appear as authentic political reporting while being generated by AI without editorial verification.

Canada's SITE Task Force made the significant step of publicly disclosing foreign interference during the active election period. The disclosure highlighted activity by a WeChat account (Youli-Youmian) linked to the Chinese Communist Party's Central Political and Legal Affairs Commission, with coordinated inauthentic behaviour and manipulated amplification tactics targeting Canadian-Chinese communities. The SITE observation indicated that AI-enhanced social engineering tools were being used by state-linked actors to target specific Canadian diaspora communities during elections.

The Canadian Centre for Cyber Security's 2025 update on cyber threats to the democratic process had assessed before the election that AI was improving the personalization and persuasiveness of social engineering attacks (Canadian Centre for Cyber Security, 2025). The election was consistent with this assessment: the combination of AI-generated images, AI-written articles, and automated bot amplification created a multi-layered disinformation environment that differed from previous Canadian elections in the simultaneous deployment of AI-generated content across multiple vectors (DFRLab, 2025; CTV News, 2025).

This record documents the broader AI-enabled election interference pattern. The specific Carney deepfake fraud campaign — which used AI to impersonate the Prime Minister for financial scams — is documented separately in a dedicated incident record.