Deepfake Video Targets Paul Vallas on Eve of Chicago Mayoral Election

On the eve of Chicago's mayoral election, a deepfake video impersonating candidate Paul Vallas was posted to Twitter, showing a fake audio of him making inflammatory statements. The video was viewed thousands of times before being taken down. The Vallas campaign condemned the video, calling it a deceptive impersonation.

Risk classification

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

The deepfake was deployed as part of a coordinated attempt to influence public opinion and manipulate a political process on the eve of an election.

Additional risk subdomains

• 4.3 Fraud, scams, and targeted manipulation: The AI system was used to impersonate a specific candidate to defame his character and manipulate voter beliefs.

Causal factors

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

The incident was caused by a human actor intentionally creating and deploying a deceptive deepfake to influence an election.

EU AI Act risk tier

• Risk tier: 3 Limited Risk

Limited Risk: The report describes an AI-generated deepfake, which falls under Risk Level 3 due to transparency obligations requiring users to be informed of its synthetic nature.

AI system and alleged parties

• AI system: None named
• AI purpose: Voice Generation; Deepfake Video Generation
• Behaviour type: Tool
• Alleged developer: Unknown deepfake creators
• Alleged deployer: Chicago Lakefront News
• Alleged harmed parties: Paul Vallas's campaign, Paul Vallas, Chicago 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 Negligible
• Differential treatment: direct Negligible, indirect Negligible
• Civil rights: direct Negligible, indirect Negligible
• Democracy: direct Minor, indirect Negligible
• Privacy: direct Negligible, indirect Negligible
• Psychological: direct Negligible, indirect Negligible
• Epistemic: direct Minor, indirect Negligible
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Malicious content

Reported: Yes, the report describes a deceptive deepfake video containing fabricated inflammatory statements about police violence.

Directly caused: The deepfake video featured a cloned voice of Paul Vallas claiming cops would kill 17 or 18 people and 'nobody would bat an eye', which was viewed thousands of times on Twitter before being removed.

Democracy

Reported: Yes, the report describes an attempt to influence a mayoral election using deceptive AI-generated media on the eve of the vote.

Directly caused: The deepfake video was posted on Twitter just before the Chicago mayoral election, potentially misleading voters and interfering with the democratic process.

Inferred additional harm: It is likely that the incident eroded trust in the integrity of the election process and political campaigns among voters who were exposed to the video.

Epistemic

Reported: Yes, the report describes the creation and spread of a highly realistic deepfake audio designed to deceive the public.

Directly caused: The video fabricated Vallas's voice saying inflammatory remarks, which was viewed thousands of times, making it difficult for normal people to distinguish real from fake.

Inferred additional harm: The incident likely contributed to a broader erosion of shared reality and increased public skepticism toward genuine campaign communications.

People affected

• Occurrences reported: 1
• People reportedly harmed: 1
• People reportedly exposed: 2000

Potential causes

Management

• Reactive Content Moderation: Platform acted after the video was reported and viewed thousands of times.

Technology

• Rapid Deepfake Generation: AI models allow generating realistic deepfakes in minutes.
• High Realism of Synthetic Audio: The deepfake audio was well made and resembled the candidate's voice.

Data Inputs

• Candidate Voice Data: Availability of candidate's public voice recordings for model training.

Human Factors

• Malicious Intent to Deceive: The creator aimed to deceive voters and damage the candidate's campaign.
• User Vulnerability to Misinformation: Normal people struggle to verify if the audio is real or fake.

Process and Methods

• Immature Detection Research: Research and tools to identify deepfakes are still in their infancy.
• Lack of Proactive Filtering: The video was uploaded and viewed thousands of times before removal.

Information quality

• Classification confidence: High
• Reason for confidence: The report clearly details the nature of the deepfake, its content, the platform used, and the immediate response. While the specific AI tool used is not named, the mechanism (voice cloning/deepfake) is well-understood.
• Ambiguities identified: The specific software or model used to generate the deepfake voice is not identified.
• Alternative interpretations: None.

A deepfake video featuring a realistic cloned voice of a Chicago mayoral candidate was deployed on the eve of an election to manipulate public opinion. While the impact was localized to a municipal election and the content was quickly removed, the incident highlights the evolving threat of AI-generated disinformation to democratic processes.

• Overall national security impact: Minor
• Scope: Single nation
• Primary target: United States
• Alleged perpetrator: Unknown

Threat characteristics

• Imminence: Immediate. The deployment occurred on the eve of an election, representing an active crisis requiring immediate mitigation within a critical 72-hour window.
• Autonomy: Human-controlled. The AI system was used strictly as a tool for voice cloning and media generation under direct human control, with no autonomous decision-making.
• Novelty: Evolved capability. Represents an evolved capability in synthetic media, demonstrating highly realistic voice cloning that is increasingly difficult for average citizens to detect.

Impact by dimension

• Physical security: Negligible. No physical security threats, kinetic attacks, or critical infrastructure compromise occurred during this incident.
• Information security: Minor. Represents an AI-enhanced disinformation campaign targeting a political candidate, but was localized to a municipal election and quickly mitigated by the platform.
• Sovereignty: Minor. Targeted a local mayoral election rather than national government functions, with no evidence of foreign state interference or disruption of constitutional processes.
• Economic security: Negligible. No significant economic or technological security implications, strategic asset theft, or financial market manipulation identified.
• Societal stability: Minor. Fabricated statements about police violence could inflame social tensions, but the rapid removal of the video limited its scale and societal impact.