This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
Russian disinformation groups, specifically Storm-1516, have utilized AI-generated content to conduct influence operations against U.S. political candidates. These operations include the creation of fraudulent campaign websites and AI-enhanced deepfake videos depicting false events, such as a staged interview claiming a candidate killed an endangered animal. These hoaxes were disseminated across multiple social media platforms to influence public opinion and undermine confidence in the democratic process.
A Russian disinformation campaign, linked to Storm-1516 and Kremlin propagandist John Mark Dougan, allegedly used AI-generated content to target Kamala Harris’s election bid. The operation included a fake campaign website promoting extreme policies and an AI-enhanced deepfake video falsely claiming Harris killed an endangered rhino in Zambia. The hoaxes were spread via Telegram, X, VK, and pro-Kremlin media outlets.
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.
The incident involves coordinated foreign influence operations using AI-generated deepfakes and fake websites to manipulate public opinion and political processes during the U.S. presidential election.
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 driven by human malicious actors who intentionally created and deployed AI-generated deepfakes and disinformation post-deployment to influence the U.S. election.
Added categories peak at 3: Substantial (Epistemic Harm, Inferred)
Russian state-backed actors utilized AI-generated deepfakes and fake campaign websites in a coordinated foreign influence operation targeting the 2024 U.S. presidential election. The campaign aimed to undermine public trust in the democratic process, representing a substantial information warfare and sovereignty concern.
The AI was used as a tool to generate and enhance specific media assets, such as deepfake videos, based on direct inputs from human creators. The reports do not describe the AI exhibiting autonomous planning, decision-making, or multi-agent coordination.
Limited Risk: The report describes the creation of 'AI-generated content such as deepfakes', which falls under Risk Level 3 due to transparency obligations to ensure users are informed of the AI's nature.
The AI systems were primarily used to generate highly realistic deepfake videos of political candidates to spread disinformation on social media.
The reports from NewsGuard and Microsoft provide detailed, consistent accounts of the foreign influence operations, identifying specific actor groups (like Storm-1516), the platforms used, and the nature of the AI-generated content (deepfakes of Kamala Harris). The role of AI as a tool for generating deceptive media is clearly established.
None. The evidence strongly points to coordinated foreign state-backed disinformation campaigns.
While the specific AI models used are not named, this information is not strictly necessary to classify the incident's domain, causal factors, or harm levels.
Technology
AI-Enhanced Deepfake Videos
Deepfakes depicted Kamala Harris making inappropriate assassination jokes.
AI-Generated Fake News Sites
Foreign actors used automation to create highly realistic spoofed domains.
Lack of Deepfake Detection Tools
Social media platforms lacked automated detection to flag AI-altered media.
Data Inputs
Spoofed Campaign Branding
Use of authentic colors and slogans to train or construct fake websites.
Staged Interview Footage
Actors and manipulated media used as inputs to feed disinformation networks.
Compromised Email Materials
Phished data from campaigns provided inputs for targeted influence operations.
Human Factors
Foreign Propagandist Manipulation
Fugitives and state-backed actors actively orchestrated and shared deepfakes.
User Cognitive Bias
Social media users shared unverified deepfakes due to confirmation bias.
Social Engineering Vulnerability
High-ranking officials fell victim to spear-phishing campaigns.
Process and Methods
Inadequate Content Moderation
Social media platforms failed to quickly remove flagged deepfake videos.
Weak Domain Registration Checks
Registrars allowed fake identities to register deceptive campaign domains.
Amplification by Troll Farms
Coordinated networks used botnets to rapidly scale AI-generated content.
Management
Inadequate Campaign Risk Assessment
Campaigns lacked robust monitoring to detect spoofed domains and deepfakes.
Platform Failure to Respond
Tech companies failed to act swiftly on reports of foreign influence assets.
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.
