Purportedly AI-Altered Images Reportedly Distort Evidence After Minneapolis Shooting of ICU Nurse Alex Pretti

Following the fatal shooting of Minneapolis ICU nurse Alex Pretti by U.S. Customs and Border Patrol agents, social media accounts reportedly circulated images purported to have been altered by AI, reportedly distorting evidence of the incident by portraying Pretti as threatening law enforcement and altering the presence of weapons. The images reportedly misidentified individuals and helped reinforce partisan narratives by purportedly obscuring verified video and eyewitness accounts.

Risk classification

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

The incident involved the coordinated use of AI-altered images by high-profile social media influencers to conduct a disinformation campaign aimed at manipulating public opinion regarding a federal shooting.

Additional risk subdomains

• 3.1 False or misleading information: The Gemini AI tool generated false details, such as removing a gun and altering faces, when used to enhance the image, creating misleading content.

Causal factors

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

The risk was driven by human actors intentionally using deployed AI tools to alter images and spread disinformation to manipulate public opinion.

EU AI Act risk tier

• Risk tier: 3 Limited Risk

Risk Level 3. Limited Risk: The report describes the use of AI-generated content and deepfakes (altered images) which pose a moderate risk and fall under transparency obligations.

AI system and alleged parties

• AI system: Gemini (Google)
• AI purpose: Image Retouching; Image Generation
• Behaviour type: Tool
• Alleged developer: Google
• Alleged deployer: Unknown disinformation actors, Unidentified social media users, Pro-Trump social media influencers, Partisan online accounts
• Alleged harmed parties: Family of Alex Pretti, Epistemic integrity, Alex Pretti

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 Minor
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Malicious content

Reported: The report explicitly describes toxic and malicious content in the form of AI-altered images and social media posts smearing the deceased victim.

Directly caused: AI-altered images were created and spread on social media, including one depicting the victim pointing a gun and another altered by Gemini that changed faces and removed a weapon.

Indirectly caused: The spread of these images fueled abusive and toxic debates online, smearing the victim as a radical activist.

Inferred additional harm: Millions of social media users (given the influencers' large follower bases of 1.4M and 3.3M) were likely exposed to this toxic and misleading content.

Democracy

Reported: The report describes how the AI-altered images were used to support partisan narratives and warp the events of a federal law enforcement shooting, which erodes public trust and democratic accountability.

Indirectly caused: The dissemination of fabricated evidence to support the administration's claims and undermine factual reporting erodes democratic oversight of law enforcement.

Inferred additional harm: Widespread use of AI to fabricate evidence in controversial state-involved shootings could severely undermine public trust in democratic institutions and the justice system.

Epistemic

Reported: The report explicitly describes epistemic harm, where AI-altered images muddled the evidence of the shooting and fueled debate over what actually happened.

Directly caused: An image processed with Gemini falsely removed a gun from an agent's hand and altered faces, while another image falsely depicted the victim pointing a gun.

Indirectly caused: The altered images muddled the factual evidence of the shooting, complicating public understanding and creating conflicting narratives online.

Inferred additional harm: Millions of social media followers were exposed to conflicting, fabricated visual evidence, eroding their ability to establish a shared consensus reality regarding the event.

People affected

• Occurrences reported: 1
• People reportedly harmed: 1
• People reportedly exposed: 4700000

Potential causes

Technology

• AI Image Hallucination: Gemini altered key details, removing a gun and changing faces in the image.
• Generative Image Manipulation: AI tools allowed users to edit a phone to look like a gun pointed at an agent.

Data Inputs

• Low Quality Source Footage: Shaky, low-resolution video inputs led to errors when AI attempted to sharpen.
• Lack of Contextual Metadata: AI tools processed images without metadata to verify real-world authenticity.

Human Factors

• Overreliance on AI Enhancements: Users trusted AI sharpening tools to reveal truth, despite inherent errors.
• Confirmation Bias in Sharing: Influencers shared AI-altered images to support pre-existing narratives.

Process and Methods

• Inadequate Verification Processes: Social media platforms lacked automated tools to flag AI-altered images.
• Lack of Provenance Tracking: No standards existed to track if images had been modified by AI tools.

Information quality

• Classification confidence: High
• Reason for confidence: The report clearly identifies the AI system (Google Gemini), its developer (Google), and the specific way it was used to alter an image of a high-profile event. The consequences of the alteration (removing a gun, changing faces) and the context of the disinformation campaign are explicitly detailed.
• Ambiguities identified: It is slightly unclear whether the image depicting the victim pointing a gun was also created using Gemini or a different AI tool, as the text separates 'One image... was edited' and 'Another was altered with Gemini'.
• Alternative interpretations: The image alterations could be viewed as simple user-driven photo editing rather than an AI safety failure, though the AI's generation of false details (removing the gun) represents a clear robustness/hallucination issue.

Following a fatal shooting by federal agents in Minneapolis, domestic social media influencers used AI-altered images to spread disinformation and support partisan narratives. While the incident highlights the evolving threat of AI-enabled epistemic harm and public opinion manipulation, its direct national security impact remains minor and manageable under standard procedures.

• Overall national security impact: Minor
• Scope: Single nation
• Primary target: United States
• Alleged perpetrator: Domestic social media influencers and users

Threat characteristics

• Imminence: Near-term. The incident represents a developing situation on social media platforms following a recent high-profile shooting.
• Autonomy: Human-controlled. The AI tools were directly operated by human users to edit and generate the altered images.
• Novelty: Evolved capability. Represents an evolved capability where AI image 'enhancement' tools are used to hallucinate and alter critical factual evidence in a shooting.

Impact by dimension

• Physical security: Negligible. No physical systems, critical infrastructure, or human safety threats were targeted or affected by the AI-altered images.
• Information security: Minor. AI-altered images were used in a domestic disinformation campaign to manipulate public perception of a federal law enforcement shooting, but lacked foreign state actor involvement.
• Sovereignty: Minor. The incident involved federal immigration agents and sparked public debate, but did not disrupt core government operations, electoral systems, or state authority.
• Economic security: Negligible. No impact on economic stability, financial systems, or strategic technological competitive advantage was reported.
• Societal stability: Minor. The AI-generated disinformation smeared a deceased victim and muddled public understanding, contributing to social polarization, though it did not trigger large-scale civil unrest.