This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
A deepfake image depicting Canadian Prime Minister Mark Carney with Jeffrey Epstein was circulated on social media platforms including X and Instagram. The image, which was used to falsely link Carney to Epstein's sex trafficking crimes, was identified as AI-generated due to visual artifacts such as misshapen hands and limbs. The content was promoted by various conservative accounts to influence public perception during a political leadership transition.
A purportedly AI-generated image that reportedly circulated on social media falsely depicted Canadian Prime Minister Mark Carney in a swimming pool with Jeffrey Epstein, implying involvement in Epstein's sex crimes. The manipulated image reportedly spread widely in early 2025, first around the time of Justin Trudeau's announced resignation, then after Carney became Liberal Party leader, and was reportedly promoted by Canadian and U.S. conservative accounts.
Domain classification, causal taxonomy, severity scores, and national security assessments were LLM-classified and may contain errors.
Using AI systems to conduct large-scale disinformation campaigns, malicious surveillance, or targeted and sophisticated automated censorship and propaganda, with the aim of manipulating political processes, public opinion, and behavior.
Also classified as
The AI-generated image was used for the targeted humiliation and defamation of Mark Carney by falsely linking him to a sex offender.
The deepfake was intentionally circulated by political accounts to influence public opinion and manipulate political processes during a leadership transition, targeting a population-level audience.
Human
Due to a decision or action made by humans
Intentional
Due to an expected outcome from pursuing a goal
Post-deployment
Occurring after the AI model has been trained and deployed
The incident was caused by human actors who intentionally generated and distributed a deepfake image to politically defame Mark Carney after the AI model was deployed.
Added categories peak at 3: Substantial (Epistemic Harm, Inferred)
An AI-generated deepfake depicting Canadian politician Mark Carney with Jeffrey Epstein was circulated during a political leadership transition. The incident represents a minor national security concern, highlighting the ongoing but manageable threat of synthetic media being used for political disinformation and character assassination.
The AI system acted as a tool to generate a synthetic image based on direct user inputs, without exhibiting autonomous planning, decision-making, or interactive assistant capabilities.
Limited Risk: The incident involves an AI system used to generate synthetic content (a deepfake), which is subject to transparency and disclosure obligations under the Act.
The AI system was used to generate a realistic but entirely fabricated image of a real person to deceive viewers.
The report provides clear, verified facts about the origin, spread, and debunking of the deepfake image, including specific social media metrics and visual artifacts.
None. The evidence clearly points to a politically motivated deepfake disinformation campaign.
Technology
AI Image Generation Flaws
Generative tools created anatomical anomalies like misshaped hands.
High-Fidelity Deepfakes
AI tools can generate highly realistic fake images of public figures.
Data Inputs
Real Photos Used as Context
A 2013 photo of Carney and Maxwell provided context for the fake image.
Human Factors
Political Confirmation Bias
Users shared the fake image to support their political narratives.
Lack of Critical Verification
Social media users accepted and shared the image without checking its source.
Process and Methods
Ineffective Platform Moderation
Social media platforms failed to detect or flag the AI-generated image.
Regulatory Environment
Lack of Deepfake Regulations
Absence of strict laws penalizing political deepfake generation.
AI system involved
An AI system was involved, but the reports do not identify it.
Identified by automated extraction; can differ from the alleged developers above.
