This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
A Conservative MP, George Freeman, was the subject of a non-consensual deepfake video that falsely depicted him announcing his defection from the Conservative Party to Reform UK. The video, which used his likeness and voice, was widely circulated on social media, causing reputational harm and raising concerns about the impact of AI-generated disinformation on democratic processes. The MP reported the incident to the police and has advocated for legislative action against political deepfakes.
A purported AI-generated deepfake video circulated online falsely depicting UK Conservative MP George Freeman announcing his defection to the Reform UK party. The fabricated clip, which mimicked Freeman’s likeness and voice without his knowledge or consent, included statements critical of the Conservatives and was widely shared on social media. Freeman denounced the video as disinformation, reported it to police, and warned that such synthetic media pose serious risks to democratic processes.
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 report also highlights the targeting of female politicians with non-consensual deepfake pornography.
The deepfake impersonated a specific MP without his consent, targeting his reputation and career.
The incident involves the creation and distribution of a political deepfake designed to mislead the public and manipulate political discourse, representing a clear case of AI-enabled disinformation.
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 a human actor intentionally creating and distributing a deepfake video post-deployment to mislead the public and damage the MP's reputation.
Added categories peak at 2: Minor (Psychological Harm, Direct)
An AI-generated deepfake video and voice clone of UK MP George Freeman falsely announced his defection to Reform UK. The incident demonstrates the rising threat of highly realistic, low-barrier AI disinformation targeting democratic processes and political figures. While causing no physical or economic harm, it highlights substantial risks to information security and democratic sovereignty, prompting electoral authorities to establish specialized monitoring capabilities.
The AI system acted as a tool to generate a synthetic video of the MP. The report notes that the 'barrier to entry' for making these videos has dropped due to 'widespread availability of AI-enabled technology', indicating it was used as a software tool by an anonymous creator.
Risk Level 3: Limited Risk. The EU AI Act classifies AI-generated content such as deepfakes under limited risk, which imposes specific transparency obligations to ensure users are informed that the content is AI-generated.
The AI system was used to generate a synthetic video and voice clone of MP George Freeman to make it appear as though he was announcing a political defection.
The reports provide consistent, detailed accounts of the specific deepfake incident targeting George Freeman, including quotes from the victim, the platforms involved, and the response of authorities. The role of AI in generating the synthetic video is clear and undisputed.
None
Technology
Widespread AI Availability
Decreased barriers to entry allow anyone to easily generate realistic deepfakes.
Realistic Voice and Image Synthesis
AI tools enable highly convincing replication of a person's voice and likeness.
Data Inputs
Unauthorised Media Usage
Existing public video and audio of the MP were used to train the AI model.
Human Factors
Anonymity of Creators
Anonymity lets creators post damaging content without immediate consequences.
Voter Susceptibility to Misinformation
Voters may easily believe and share realistic-looking deepfakes on social media.
Process and Methods
Inadequate Platform Moderation
Meta and other platforms failed to quickly flag and remove the deepfake video.
Viral Social Media Distribution
Algorithms and sharing mechanisms allowed the video to spread rapidly online.
Management
Inadequate Platform Guidelines
Platform policies did not classify the political deepfake as a guideline breach.
Regulatory Environment
Lack of Specific Laws
No clear laws exist targeting the unauthorized use of identity in deepfakes.
Lagging Legal Frameworks
The law has not kept pace with rapid developments in AI-enabled threats.
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.
