This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
In April 2022, Northern Irish politician Cara Hunter was targeted by a deepfake pornography campaign weeks before an assembly election. The fake video, which falsely depicted her in explicit acts, was circulated widely on WhatsApp, leading to severe real-world harassment, including street intimidation and reputational damage. The incident highlights the use of AI-generated synthetic media as a tool for political sabotage and gender-based violence, with the victim noting that the lack of specific legislation and cyber-crime resources left her without legal recourse.
Sometime in April 2022, reportedly some weeks before the May 5, 2022 Northern Ireland Assembly election, Member of the Legislative Assembly Cara Hunter was targeted in a deepfake pornography incident. A fake video depicting her in explicit scenarios is reported to have circulated widely on WhatsApp. Law enforcement have as of yet been unable to trace the origins of the video, according to reports. Additionally, it is alleged that about six months after the initial incident, she was the target of at least 15 more AI-generated deepfakes depicting her likeness in underwear.
Domain classification, causal taxonomy, severity scores, and national security assessments were LLM-classified and may contain errors.
Using AI systems to gain a personal advantage over others such as through cheating, fraud, scams, blackmail or targeted manipulation of beliefs or behavior. Examples include AI-facilitated plagiarism for research or education, impersonating a trusted or fake individual for illegitimate financial benefit, or creating humiliating or sexual imagery.
Also classified as
The incident exposed the victim, her family, and the public to highly explicit, non-consensual pornographic synthetic media.
The deepfake video generated and spread false information about the victim's actions, leading to inaccurate beliefs among voters.
The incident involved the creation and viral distribution of non-consensual sexual deepfakes to target, humiliate, and politically manipulate the reputation of a specific victim.
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 initiated by a malicious human actor who intentionally used a deployed AI deepfake tool to generate and distribute explicit content targeting the victim.
Added categories peak at 3: Substantial (Epistemic Harm, Inferred)
A local Northern Irish politician was targeted with a non-consensual deepfake pornography campaign weeks before an election, leading to harassment and reputational damage. The incident highlights the growing threat of AI-enabled political sabotage and gender-based harassment, though its impact remained localized and did not escalate to a systemic national security crisis.
The AI system was used as a tool to perform a specific, well-defined function of superimposing a face onto an explicit video. It operated on direct input without autonomous decision-making or planning capabilities.
Limited Risk: The EU AI Act classifies AI-generated content such as deepfakes under Limited Risk, which imposes specific transparency obligations such as notifying users that the content is AI-generated.
The AI system was used to generate a highly realistic, explicit video by superimposing the victim's face onto another body, as well as subsequent lingerie images.
The reports provide consistent, detailed first-hand accounts from the primary victim, Cara Hunter, regarding the timing, dissemination, and real-world impacts of the deepfake video. The context of the 2022 Northern Ireland Assembly election and the subsequent legislative efforts are well-documented across multiple reputable news sources.
None. The reports consistently identify the video as a non-consensual deepfake used for targeted political and sexual harassment.
Technology
Rapid AI face-swap advancement
AI makes creating highly realistic face-swaps fast and accessible to anyone.
Accessible nudification apps
Cheap or free apps allow users to virtually undress any photo with ease.
Lack of AI watermarking
Absence of mandatory markers on synthetic media hinders verification.
Data Inputs
Stolen social media photos
Publicly available images are harvested to train deepfake models.
Innocent source videos
Ordinary internet videos are used to map faces onto explicit acts.
Human Factors
Targeted political misogyny
Deepfakes are weaponized to humiliate and silence female politicians.
Public apathy to deepfakes
Some people view non-consensual deepfakes as harmless entertainment.
Victim shaming and isolation
Social stigma leads to victims being shunned in their communities.
Process and Methods
Slow platform moderation
Social media networks fail to quickly remove explicit synthetic content.
Encrypted viral distribution
WhatsApp encryption allows rapid, anonymous, and untraceable sharing.
Management
Inadequate cyber-crime funding
Police departments lack the technology and expertise to trace deepfakes.
Prioritizing growth over safety
Social media giants fail to implement proactive deepfake protections.
Lack of candidate preparation
Electoral bodies do not prepare candidates for digital smear campaigns.
Regulatory Environment
Legislative gaps on creation
Existing laws target distribution but fail to ban deepfake creation.
Slow legislative processes
Government regulations cannot keep pace with rapid AI advancements.
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.
