This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
A 40-year-old man used ChatGPT to generate a fake image of an escaped wolf named Neukgu, which he then shared in a group chat. The image was mistakenly used by Daejeon authorities in official alerts and press briefings, leading to a diversion of search resources. The man is now under investigation for disrupting government work by deception.
South Korean police alleged that a 40-year-old man used ChatGPT's image-generation capabilities to add escaped wolf Neukgu to a Daejeon street photo, then posted the purportedly fabricated image in a company group chat. Authorities reportedly used the image in alerts and briefings, redirecting search resources and disrupting public-safety operations before police identified him through CCTV and AI-use records.
Domain classification, causal taxonomy, severity scores, and national security assessments were LLM-classified and may contain errors.
AI systems that inadvertently generate or spread incorrect or deceptive information, which can lead to inaccurate beliefs in users and undermine their autonomy. Humans that make decisions based on false beliefs can experience physical, emotional or material harms
Also classified as
Government authorities over-relied on the unverified AI-generated image, using it to redirect search operations and issue official emergency alerts without verifying its authenticity.
The incident involved the generation of a fake image that misled both the public and government authorities, leading to inaccurate beliefs and the diversion of public safety resources.
Human
Due to a decision or action made by humans
Unintentional
Due to an unexpected outcome from pursuing a goal
Post-deployment
Occurring after the AI model has been trained and deployed
The incident was caused by a human decision to generate and share a fake image. The disruption of government work was an unintentional consequence of the user's action, which he claimed was done 'for fun'.
Added categories peak at 2: Minor (Epistemic Harm, Direct)
An individual used ChatGPT to generate a fake image of an escaped wolf, which municipal authorities in South Korea mistakenly used to redirect search operations and issue public emergency alerts. While the incident caused localized operational disruption and public concern, it represents a minor national security impact, highlighting vulnerabilities in government verification of AI-generated media.
The AI system, ChatGPT, was used as a tool by the human user to perform a specific, well-defined function of generating an image based on direct text prompts, without autonomous planning or decision-making by the AI itself.
Limited Risk: The system is a general-purpose AI used to generate synthetic content (an AI-generated image), which falls under transparency obligations for AI-generated content and deepfakes.
The AI system was used to generate a realistic but fake image of an escaped wolf to mislead others, which was then shared in a group chat.
The report provides clear, consistent details about the incident, including the AI system used, the actions of the suspect, the reaction of the authorities, and the legal consequences. There are no major contradictions.
None
Technology
Realistic AI Generation
Software enabled creation of highly realistic fake image of the wolf.
Accessible AI Tools
Easy access to AI tools allowed rapid creation of misleading content.
Data Inputs
Lack of Synthetic Metadata
No digital metadata or watermarks identified the image as AI-generated.
Human Factors
Malicious or Careless Creation
Individual created and shared the fake image 'for fun' without caution.
Cognitive Bias of Authorities
Search teams rushed to believe the realistic image without verification.
Process and Methods
Inadequate Source Verification
No system in place to verify the origin of images found online.
Flawed Emergency Alert Protocol
City issued emergency texts based on unverified social media images.
Management
Poor Crisis Information Management
Failure to establish clear verification steps before acting on tips.
Regulatory Environment
Weak AI Safety Regulations
Lack of strict rules preventing the misuse of generative AI tools.
AI system involved
Identified by automated extraction; can differ from the alleged developers above.
