Asked AI About School Attack Plan, Caught by U.S. FBI

Conversations with ChatGPT About School Attack and Murder Plot
Digital Devices Seized, Forensic Analysis Underway

In Argentina, a 15-year-old teenager was caught by the local branch of the United States Federal Bureau of Investigation (FBI) after plotting with artificial intelligence (AI) service ChatGPT to attack a school and kill classmates. The teenager is now under investigation.

ChatGPT. AP Yonhap News

On September 11 (local time), local daily newspaper La Nacion and other foreign media reported that the Argentine Federal Police launched an investigation after receiving information about the case from a legal attaché of the FBI at the U.S. Embassy in Argentina.

Based on the conversation the teenager had with ChatGPT, the police identified the 15-year-old residing in Quilmes, Buenos Aires Province. A search of the suspect’s home led to the discovery of digital devices containing data related to plans for a school attack and the murder of classmates. Specifically, authorities said the teenager discussed an attack at the school to take place next year, and investigators confirmed that the suspect even mentioned to ChatGPT the possibility of moving the attack date forward.

During the raid, police seized electronic devices including a mobile phone. Through digital forensic analysis, they aim to determine the extent to which actual preparations had been made for the crime, and to see if other students or third parties were involved in the plan. However, no firearms or ammunition were found during the search.

The investigation is underway on charges of "threats against the public." The teenager and the parents are also expected to be questioned by the prosecution and the courts.

Meanwhile, on August 14 in Florida, USA, Darren Zhou (25) was arrested after engaging in a conversation with ChatGPT about "killing his ex-girlfriend." Zhou reportedly outlined detailed plans to ChatGPT, stating, "I’ll kill her by the end of this month" and that he would wait for his ex-girlfriend with flowers at the gym parking lot where she works out, and if rejected, kill her with a gun and then end his own life. OpenAI detected Zhou’s harmful conversational patterns and reported the case to the FBI based on this information.

Previously, OpenAI became embroiled in a lawsuit after failing to report to relevant authorities that the perpetrator of a mass shooting in British Columbia, Canada, had repeatedly outlined concrete scenarios for the crime to ChatGPT several months prior to the attack. At the time, OpenAI CEO Sam Altman stressed, “We are committed to finding ways to ensure such tragedies never happen again.”

This content was produced with the assistance of AI translation services.