This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
A 16-year-old in Tira, Israel, allegedly consulted ChatGPT to explore attack methods before attempting to stab a police officer at a police station on March 5, 2025. The teenager was charged with attempted aggravated assault as an act of terror. The extent to which ChatGPT's responses influenced the attack is not established.
A 16-year-old from Tira, Israel, allegedly entered the city's police station with a knife and attempted to stab a Border Police officer. Prosecutors reportedly alleged that before the attack he had consulted ChatGPT about possible attacks and courses of action. He was charged in juvenile court with attempted aggravated assault as an act of terror. Public reporting does not establish what responses ChatGPT provided or whether they materially assisted the attempted attack.
Domain classification, causal taxonomy, severity scores, and national security assessments were LLM-classified and may contain errors.
AI that exposes users to harmful, abusive, unsafe or inappropriate content. May involve providing advice or encouraging action. Examples of toxic content include hate speech, violence, extremism, illegal acts, or child sexual abuse material, as well as content that violates community norms such as profanity, inflammatory political speech, or pornography.
Also classified as
The AI system's safety filters failed to prevent the user from obtaining instructions on executing a violent attack.
The AI system allegedly provided the user with methods and tactics for carrying out a terrorist attack, exposing them to toxic and unsafe content regarding violence and illegal acts.
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 the intentional actions of a human user who sought out the AI to plan a violent attack post-deployment.
Added categories peak at 2: Minor (Psychological Harm, Inferred)
A teenager in Israel consulted ChatGPT to research tactics before attempting a knife attack on a police officer. The attack was foiled without injury. While the incident demonstrates the risk of AI being used to assist in terrorist planning, the overall national security impact is minor, representing a human-controlled threat with localized consequences.
The teenager used ChatGPT as an assistant to 'explore possible methods and tactics for carrying out a terrorist act'. The AI acted to provide information and suggestions to augment the user's planning, requiring human decision-making to execute.
Risk Level 3. Limited Risk: The system involved is ChatGPT, which is a general-purpose chatbot. Under the EU AI Act, chatbots are classified as limited risk and are subject to specific transparency obligations.
The AI system used was ChatGPT, a general-purpose conversational chatbot designed to answer user queries and assist with text generation.
The reports clearly state that the teenager consulted ChatGPT to plan the attack, but they lack specific details on what prompts were used, what exact information the AI provided, or how much the AI's output actually influenced the final attack plan. The indictment details are official, which increases credibility, but the technical role of the AI is described only at a high level.
The incident could be viewed purely as a human-driven terrorist attempt where the AI's role was negligible or acted merely as a standard search engine replacement, rather than an AI-specific failure.
The specific prompts entered by the user, the exact text of the responses generated by ChatGPT, and any forensic analysis showing how the AI's advice was incorporated into the attack plan.
Technology
Flawed Safety Guardrails
The chatbot failed to block queries about planning a terrorist attack.
Lack of Real-Time Detection
No automated system flagged the user's dangerous intent to authorities.
Data Inputs
Harmful Training Information
Model was trained on data containing tactical planning and attack methods.
Unfiltered User Prompts
Input prompts asking for attack tactics were processed without blocking.
Human Factors
Malicious User Intent
The radicalized teen actively sought instructions to conduct an attack.
Process and Methods
Weak Alignment Protocols
Reinforcement learning failed to prevent the generation of harmful advice.
Management
Insufficient Abuse Monitoring
Management failed to implement proactive monitoring for high-risk users.
Regulatory Environment
Absence of AI Safety Mandates
No legal frameworks forced the developer to block tactical planning queries.
AI system involved
Identified by automated extraction; can differ from the alleged developers above.
