AI deepfakes of dozens of Canadian women in violent and sexual images shared online

✕ Headline / Body Mismatch: The headline accurately summarizes the core event (AI deepfakes of Canadian women shared online) without exaggeration. It avoids sensationalist language while conveying the seriousness of the issue.

"AI deepfakes of dozens of Canadian women in violent and sexual images shared online"

✕ Appeal to Emotion: The article uses emotionally resonant language from victims (e.g., 'violated', 'terrified') but attributes them clearly, maintaining objectivity. The reporter does not insert personal judgment.

"Multiple women across Canada say they feel confused, violated and terrified after learning pictures of themselves were allegedly taken from their social media profiles..."

✕ Loaded Language: The phrase 'utterly disgusting' appears in a victim quote, not in the reporter’s voice, preserving neutrality in tone.

"“It was disgusting to see that like an innocent photo of myself, he had turned to something so vile,” said one woman..."

✕ Loaded Labels: The article avoids loaded labels like 'pervert' or 'monster' and refrains from editorializing the accused, using neutral descriptors like 'charged' and 'alleged'.

"Stephen Lowe, 60, of Maitland, N.S., is facing 79 charges including..."

✕ Passive-Voice Agency Obfuscation: Passive voice is used appropriately when agency is unknown (e.g., 'photos were posted'), but active voice is used when known (e.g., 'Lowe was arrested'), preserving clarity of responsibility.

"Lowe was arrested on Feb. 12 by the RCMP in Nova Scotia."

✕ Scare Quotes: The article includes graphic descriptions of content (rape, torture) because they are central to the crime, but does so with restraint and purpose, not for shock value.

"The women who spoke to CBC News describe violent and graphic imagery that includes their real faces in AI-generated scenes, including rape, torture, bondage, and kidnapping."

✓ Comprehensive Sourcing: The article includes multiple named victims (using pseudonyms for protection), law enforcement sources, legal experts, family members, and references court documents—ensuring diverse and credible sourcing across affected parties and institutions.

"Suzie Dunn, director of the law and technology institute at Dalhousie University and an expert in technology-facilitated violence with a focus on intimate images, says she and others have been calling for legal responses to deepfakes for nearly a decade and that Canada is “far behind.”"

✓ Viewpoint Diversity: Victims are quoted directly and given space to describe emotional and psychological impacts, ensuring their voices are central without being editorialized.

"“This is never going to go away,” she said. “I'm always going to worry about my body being exploited and shared.”"

✓ Proper Attribution: The accused are not quoted directly, but their lawyers’ positions (declining to comment) are accurately reported, and the article consistently notes that allegations are unproven—maintaining fairness in sourcing despite the gravity of claims.

"These allegations have not been proven in court. Lawyers for both accused declined to comment."

✓ Viewpoint Diversity: The wife of one accused is given a platform to express shock and condemnation, adding emotional depth and avoiding a one-dimensional portrayal of personal connections.

"I can say unequivocally that my heart is forever broken for the women impacted by these charges."

✕ Framing by Emphasis: The article frames the story around victim impact and systemic failure in law enforcement and legislation, rather than reducing it to a crime narrative or political debate. This is a legitimate and ethically sound framing given the subject.

"“This is never going to go away,” she said. “I'm always going to worry about my body being exploited and shared.”"

✕ Narrative Framing: The narrative includes calls for legal reform and expert commentary, elevating it beyond episodic reporting to address structural issues—avoiding isolation of the event as a singular scandal.

"Canada’s justice minister spoke out about the matter after a Halifax man accused of creating deepfakes of women he knew was acquitted in March."

✕ Framing by Emphasis: The story avoids conflict framing between opposing political views and instead focuses on consensus among victims, experts, and even the accused’s wife on the need for stronger laws.

"T.T., along with other alleged victims, Lowe’s wife and experts, are calling on the federal government to strengthen Canada’s laws around the creation of this kind of content."

✓ Contextualisation: The article provides historical context about Canada's existing laws against non-consensual intimate images since 2014 and explains the legal gap regarding deepfakes. It also references prior judicial criticism and current legislative efforts (Bill C-16), giving readers a clear timeline and policy background.

"It has been illegal to share intimate images without consent in Canada since 2014, but experts say there is a gap in the law such that it doesn’t include deepfakes."

✓ Contextualisation: The piece includes international comparative context by citing Australia’s 2024 criminalization of deepfakes, helping readers understand Canada’s relative position in global policy development.

"She points to Australia, which criminalized deepfakes in 2024, as a leader in this area."

✓ Contextualisation: The article notes that charges relate to a specific timeframe (Aug 1, 2020–Dec 30, 2025), but some victims found content posted earlier—highlighting potential limitations in the current legal scope or investigation breadth.

"The charges currently before the courts for Lowe pertain to the time frame between Aug. 1, 2020 and Dec. 30, 2025, but some of the women said they found photos of themselves that were posted many years prior."