This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
President Donald Trump posted a deepfake video on Truth Social depicting Senate Minority Leader Chuck Schumer and House Minority Leader Hakeem Jeffries in a derogatory manner. The video, which used AI to alter the subjects' appearance and speech, was condemned by Democratic leaders as racist and a distraction from critical government shutdown negotiations. The specific AI tool used to create the video remains unidentified.
President Donald Trump reportedly posted a purportedly AI-modified video on Truth Social depicting Senate Minority Leader Chuck Schumer making fabricated statements and House Minority Leader Hakeem Jeffries in a sombrero. Critics, including Jeffries and Rep. Ro Khanna, reportedly condemned the video as racist and misleading during budget negotiations over a looming government shutdown. The AI tool used for modification has not been identified.
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 video depicted Hakeem Jeffries in a bigoted, racially stereotypical caricature using a sombrero and handlebar mustache.
The video targeted specific individuals, Schumer and Jeffries, using synthetic media to humiliate and misrepresent them.
The deepfake video was used by a prominent political figure to influence public opinion and political processes during critical government shutdown negotiations.
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 posting a deepfake video using a deployed AI tool to mock political opponents.
Added categories peak at 2: Minor (Epistemic Harm, Direct)
President Donald Trump posted a deepfake video on Truth Social depicting Democratic leaders Chuck Schumer and Hakeem Jeffries in a derogatory and racially caricatured manner during active government shutdown negotiations. While the incident represents a minor distraction to government functions and highlights the ongoing challenge of political synthetic media, it has negligible impact on physical, economic, or intelligence security.
The AI system was used as a tool to digitally alter the video and audio of the politicians based on direct inputs provided by the creator, without exhibiting autonomous planning or interactive assistant behaviors.
Risk Level 3. Limited Risk: AI-generated content such as deepfakes. The report describes an AI-modified deepfake video used to alter the appearance and speech of public figures.
The AI system was used to generate a deepfake video altering the appearance of Hakeem Jeffries and Chuck Schumer, and to generate synthetic voice/audio for Schumer.
The report clearly describes the incident, the platform where it was posted, the specific content of the deepfake, and the reactions of the affected parties. While the specific AI tool is not named, the nature of the AI involvement (deepfake video/audio generation) is unambiguous.
The incident could be viewed primarily as a political speech/satire issue rather than an AI safety incident, though the use of synthetic media to manipulate public perception makes it a clear AI safety case.
Technology
AI Video Modification Tools
Sophisticated AI tools allowed realistic alteration of speech and appearance.
Ease of Deepfake Creation
Accessible AI technology enables rapid generation of convincing political fakes.
Data Inputs
Source Video Exploitation
Authentic footage of public figures was used as input for AI alteration.
Human Factors
Malicious Political Intent
Using deepfakes intentionally to mock opponents and spread misinformation.
Audience Polarization
Partisan audiences may accept deepfakes as true or acceptable political jest.
Process and Methods
Lack of Content Labeling
No automated process existed to label the video as AI-generated on posting.
Management
Permissive Platform Moderation
Truth Social management allowed the hosting and distribution of deepfakes.
Regulatory Environment
Inadequate AI Regulations
Lack of legal frameworks prohibiting deepfakes in political campaigns.
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.
