Texas man arrested after AI chatbot flags school shooting threats

Nathaniel Michael Carrasco, a 22-year-old man from San Antonio, Texas, has been arrested following threats made to an AI chatbot regarding a planned attack on a local elementary school. According to investigators, Carrasco used Anthropic’s Claude chatbot to express intentions to obtain a firearm and “shoot up kids” at Serna Elementary School, which is located adjacent to his residence.

During the interactions, Carrasco allegedly requested images of “dead school kids bleeding” and explicitly stated, “I truly do mean my threats.” Anthropic’s safety protocols detected the violent content and alerted the FBI’s National Threat Operations Center. Following the alert, the FBI and San Antonio police worked with technology and telecommunications providers to identify and locate the suspect.

Carrasco is currently being held in Bexar County on charges related to terrorist threats, with bail set at $250,000. The incident has raised concerns regarding school security and the effectiveness of AI safety guardrails in preventing real-world violence.