Suspect in Palisades Fire Allegedly Consulted ChatGPT for Arson Tips and Legal Advice Before Blaze That Killed 12 and Destroyed 6,837 Structures

Federal prosecutors allege that Jonathan Rinderknecht, arrested on 10/08/2025, consulted ChatGPT for arson-related imagery and legal advice before and after deliberately starting the Palisades fire in Los Angeles on 01/01/2025. The blaze, which reignited on 01/07/2025, killed 12 people and destroyed 6,873 buildings. Prosecutors say he generated a dystopian fire image with ChatGPT and later asked about liability for fires caused by cigarettes.

Risk classification

• Primary risk domain: 1 Discrimination & Toxicity
• Primary risk subdomain: 1.2 Exposure to toxic content

The suspect used ChatGPT to generate content related to illegal acts, specifically visualizing a burning city and seeking advice on legal liability for starting a fire.

Additional risk subdomains

• 7.3 Lack of capability or robustness: The AI system failed to refuse prompts that visualized mass destruction and queried legal liability for starting a fire, showing a lack of robust safety filters.

Causal factors

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

The catastrophic wildfire and subsequent harm were caused by the intentional physical actions of a human suspect who set the fire, rather than an autonomous decision by the AI.

EU AI Act risk tier

• Risk tier: 3 Limited Risk

Limited Risk: ChatGPT is a general-purpose chatbot and image generator, which falls under Risk Level 3 due to transparency obligations for AI systems interacting with humans and generating content.

AI system and alleged parties

• AI system: ChatGPT (OpenAI)
• AI purpose: Chatbot; Image Generation
• Behaviour type: Assistant
• Alleged developer: OpenAI
• Alleged deployer: Jonathan Rinderknecht
• Alleged harmed parties: Victims of the Palisades Fire, Pacific Palisades community, General public of Los Angeles, General public of California

Harm severity

Severity ratings are based on this scale.

Highest direct severity in any category: Negligible. 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 Substantial
• Infrastructure: direct Negligible, indirect Substantial
• Property: direct Negligible, indirect Severe
• Financial: direct Negligible, indirect Negligible
• Environmental: direct Negligible, indirect Substantial
• Malicious content: direct Negligible, indirect Negligible
• Differential treatment: direct Negligible, indirect Negligible
• Civil rights: direct Negligible, indirect Negligible
• Democracy: direct Negligible, indirect Negligible
• Privacy: direct Negligible, indirect Negligible
• Psychological: direct Negligible, indirect Negligible
• Epistemic: direct Negligible, indirect Negligible
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Physical

Reported: The report explicitly describes physical harm, stating that 12 people were killed in the Palisades fire.

Indirectly caused: The Palisades fire, which the suspect researched and visualized using ChatGPT, resulted in 12 fatalities.

Inferred additional harm: It is highly likely that numerous other individuals suffered non-fatal physical injuries, smoke inhalation, or burns from the massive wildfire, though specific injury numbers are not detailed in the report.

Infrastructure

Reported: The report explicitly describes infrastructure damage, noting that thousands of buildings were destroyed.

Indirectly caused: The fire destroyed 6,873 buildings, which likely included residential structures and local utility infrastructure in the affected hiking trail areas.

Inferred additional harm: Damage to local utility grids, communication networks, and public roads in the Los Angeles west side is highly likely given the scale of the fire.

Property

Reported: The report explicitly describes property damage, stating that 6,873 buildings were destroyed.

Indirectly caused: The Palisades fire destroyed 6,873 buildings, including residential structures and hiking trail facilities.

Inferred additional harm: The destruction of thousands of homes and personal belongings represents billions of dollars in property damage.

Environmental

Reported: The report explicitly describes environmental damage, noting that the fire tore across federal land with hiking trails and burned vegetation.

Indirectly caused: The fire burned federal land, hiking trails, and extensive vegetation across Los Angeles' west side.

Inferred additional harm: Massive destruction of local ecosystems, wildlife habitats, and air quality degradation from smoke across the Los Angeles basin.

People affected

• Occurrences reported: 1
• People reportedly harmed: 12
• People reportedly exposed: 15000

Potential causes

Technology

• Permissive Image Generation: ChatGPT generated dystopian image of a burning forest without safety block.
• Inadequate Query Filtering: AI allowed queries regarding legal fault for starting fires.

Data Inputs

• Lack of Contextual Input Data: AI lacked real-time context of the user's physical actions or intent.

Human Factors

• User Malicious Intent: The user intentionally used AI to visualize and rationalize arson.

Process and Methods

• Weak Safety Guardrails: Safety guardrails failed to flag prompts visualizing destructive fires.

Information quality

• Classification confidence: High
• Reason for confidence: The reports and extraction context provide clear details about the suspect's actions, the specific prompts used, and the resulting fire damage.
• Ambiguities identified: The exact degree to which the ChatGPT outputs influenced or encouraged the suspect's decision to commit arson remains legally and psychologically ambiguous.
• Alternative interpretations: The incident could be viewed purely as a human criminal act where the AI played an incidental, non-causal role, rather than an AI safety failure.

A suspect arrested for a catastrophic arson in Los Angeles used ChatGPT to generate imagery of a burning city and query legal liability beforehand. While the physical consequences of the fire were severe, the AI's role was strictly auxiliary and human-controlled, representing a minor national security impact managed under standard law enforcement procedures.

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

Threat characteristics

• Imminence: Long-term. The suspect has been arrested and the active crisis is resolved, representing an ongoing policy and safety-filter concern rather than an active threat.
• Autonomy: Human-controlled. The AI functioned purely as an assistant, generating text and images based on direct user prompts, with no autonomous decision-making.
• Novelty: Evolved capability. Represents an advancement where criminals use generative AI to visualize crimes and query legal liability prior to execution, evolving from traditional search engine queries.

Impact by dimension

• Physical security: Minor. The suspect committed a devastating physical arson causing 12 deaths, but the AI's role was limited to generating pre-incident imagery and a basic liability query, rather than executing or directing the attack.
• Information security: Negligible. No information warfare, intelligence compromise, or systematic disinformation campaigns were associated with this incident.
• Sovereignty: Negligible. Core government operations, sovereignty, and electoral systems were not impacted by the suspect's use of the AI.
• Economic security: Negligible. No strategic technology theft, financial system manipulation, or threat to technological competitive advantage occurred.
• Societal stability: Negligible. While the fire caused local tragedy and displacement, the AI system itself did not drive mass surveillance, algorithmic oppression, or systemic civil rights violations.