AI-Generated Images of Trump with Black Voters Spread as Disinformation Before U.S. Primary Elections

In the run-up to the U.S. primary elections, supporters of Donald Trump shared AI-generated images showing him with Black voters in an attempt to sway African-American votes. These deepfakes, including Trump's distorted hand visuals, were initially created by satirical accounts but were later misappropriated for political disinformation, misleading millions on social media platforms.

Risk classification

• Primary risk domain: 4 Malicious actors
• Primary risk subdomain: 4.1 Disinformation, surveillance, and influence at scale

The incident involves the coordinated or at-scale dissemination of AI-generated political deepfakes to influence public opinion and manipulate democratic processes among African-American voters.

Additional risk subdomains

• 3.1 False or misleading information: The AI-generated images directly fabricated false events (e.g., Trump posing with specific voters, Trump praying in a church) that misled some social media users.

Causal factors

• Entity: Human
• Intent: Intentional
• Timing: Post-deployment

The risk was primarily caused by human supporters intentionally co-opting and sharing AI-generated images with false captions to mislead voters after the AI models were deployed.

EU AI Act risk tier

• Risk tier: 3 Limited Risk

Limited Risk: The report describes the generation and dissemination of deepfakes, which are AI-generated content subject to specific transparency obligations under the Act.

AI system and alleged parties

• AI system: unspecified
• AI purpose: Image Generation; Deepfake Image Generation
• Behaviour type: Tool
• Alleged developer: Various social media accounts, Trump supporters
• Alleged deployer: Various social media accounts, Trump supporters
• Alleged harmed parties: Public discourse integrity, General public, Democracy, Black voters, African-American voters

Harm severity

Severity ratings are based on this scale.

Highest direct severity in any category: Minor. Severity is scored from Negligible to Catastrophic in each harm category, for harm the reports describe as caused directly or indirectly by the AI system.

• Physical: direct Negligible, indirect Negligible
• Infrastructure: direct Negligible, indirect Negligible
• Property: direct Negligible, indirect Negligible
• Financial: direct Negligible, indirect Negligible
• Environmental: direct Negligible, indirect Negligible
• Malicious content: direct Minor, indirect Minor
• Differential treatment: direct Negligible, indirect Negligible
• Civil rights: direct Negligible, indirect Negligible
• Democracy: direct Negligible, indirect Minor
• Privacy: direct Negligible, indirect Negligible
• Psychological: direct Negligible, indirect Negligible
• Epistemic: direct Minor, indirect Severe
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Malicious content

Reported: The report explicitly describes the creation and spread of malicious synthetic content (deepfakes) designed to mislead voters.

Directly caused: AI image generators were used to create realistic but fabricated images of Donald Trump with Black voters and in other misleading scenarios.

Indirectly caused: Supporters shared these deepfakes on social media platforms like Twitter/X, where one image gained over 1.4 million views, exposing a large audience to deceptive content.

Inferred additional harm: It is likely that many more political deepfakes were generated and circulated across various platforms, exposing millions of voters to deceptive synthetic media.

Democracy

Reported: The report explicitly describes the erosion of democratic norms through the use of fake photographs as a powerful political disinformation tool.

Indirectly caused: The dissemination of deepfakes with false captions aimed to manipulate African-American voters, undermining the integrity of democratic election campaigns.

Inferred additional harm: The widespread, unregulated use of political deepfakes is highly likely to erode public trust in democratic processes and election integrity at a national scale.

Epistemic

Reported: The report explicitly describes epistemic harm through the generation and spread of false images that misled some social media users.

Directly caused: AI tools generated highly realistic images of Trump in fabricated scenarios, which lacked obvious signs of manipulation to casual viewers.

Indirectly caused: A supporter reposted an image with a false caption, leading to over 1.4 million views on Twitter/X, where some users believed the event actually occurred.

Inferred additional harm: The proliferation of high-quality political deepfakes makes it increasingly difficult for the public to distinguish between real and synthetic events, eroding shared consensus reality.

People affected

• Occurrences reported: 1
• People reportedly exposed: 1400000

Potential causes

Management

• Campaign Use of Unverified Media: Political figures shared unverified AI images to boost their public image.

Technology

• Generative Text-to-Image Software: Enabled easy creation of highly realistic but fabricated images of Trump.
• Anomalous Image Artifacts: AI struggled with rendering realistic hands, creating telltale distortions.

Data Inputs

• Lack of Image Provenance Data: Images lacked metadata or watermarks to verify their synthetic origin.

Human Factors

• Malicious Co-optation of Satire: Supporters repurposed satirical AI images with false, misleading captions.
• Public Susceptibility to Fake Media: Some users believed the realistic deepfakes were actual photographs.

Process and Methods

• Viral Social Media Distribution: Platforms allowed rapid sharing, gaining millions of views without checks.
• Inadequate Platform Moderation: Social media failed to flag or remove misleading AI-generated political media.

Regulatory Environment

• Absence of Deepfake Regulations: Lack of legal frameworks to govern political AI-generated disinformation.

Information quality

• Classification confidence: High
• Reason for confidence: The report provides clear details about the creation, dissemination, and social media reach of the AI-generated images, as well as the political context. The role of the AI as an image-generation tool is clear, and the human intent to mislead is well-documented.
• Ambiguities identified: The specific AI models used to generate the images are not identified.

This incident highlights the growing national security risk of AI-generated synthetic media being used for political disinformation. Supporters co-opted realistic AI images of Donald Trump with Black voters to influence democratic processes, demonstrating how easily accessible generative tools can scale targeted influence operations during critical election cycles.

• Overall national security impact: Substantial
• Scope: Single nation
• Primary target: United States
• Alleged perpetrator: Supporters of Donald Trump

Threat characteristics

• Imminence: Near-term. The threat of AI-generated political disinformation is an active, developing concern throughout the ongoing election cycle.
• Autonomy: Human-controlled. The AI systems used were text-to-image generators operating under direct human control, with humans generating, co-opting, and distributing the images.
• Novelty: Evolved capability. While political disinformation is established, the ease of generating highly realistic synthetic media targeting specific voter demographics represents an evolved and highly accessible capability.

Impact by dimension

• Physical security: Negligible. No physical security threats, kinetic targeting, or critical infrastructure manipulation were reported in this incident.
• Information security: Substantial. AI-generated deepfakes were used in an election disinformation campaign targeting Black voters, gaining millions of views and demonstrating the scale of synthetic media in political influence operations.
• Sovereignty: Minor. The incident involves attempts to manipulate voter perceptions during democratic elections, though no foreign state actor involvement or direct compromise of electoral infrastructure was reported.
• Economic security: Negligible. No threats to economic stability, strategic technology theft, or critical supply chain vulnerabilities were indicated.
• Societal stability: Minor. The campaign targeted specific demographic groups to manipulate public opinion, contributing to epistemic harm and potential erosion of trust in public information, but did not cause civil unrest.