A 22-year-old San Antonio man who lives next door to an elementary school is facing a felony charge after investigators say he told an artificial intelligence chatbot he planned to shoot students there once classes began. Nathaniel Michael Carrasco was arrested on a third-degree felony charge of making a terroristic threat and remained in the Bexar County jail in lieu of $250,000 bail.
According to an arrest affidavit cited by the San Antonio Express-News, Carrasco made the threats in queries to an Anthropic AI chatbot around August 11, days before the North East Independent School District was scheduled to begin the school year. Carrasco allegedly told the chatbot: “I'm getting a gun as soon as I can, and when Serna Elementary starts, I'm just gonna walk in there and start shooting up kids.” He also reportedly wrote, “I hope I can get a gun so that way I can just shoot up a school already,” and used the chatbot to search for “dead school kids bleeding.”
Serna Elementary sits next door to Carrasco's home near Loop 410 and Perrin Beitel Road on San Antonio's Northeast Side, per the affidavit as reported by the Express-News. The campus serves roughly 341 Pre-K through 5th-grade students within the North East Independent School District, according to NEISD, and sits adjacent to residential neighborhoods in the area.
Chatbot Messages Described as “No Escape”
The affidavit quotes additional messages attributed to Carrasco, including “There shall be no escape, no matter what,” and “I truly do mean my threats.” He also reportedly said he would “do everything in his power to help create that reality.” The document does not say how the AI chatbot responded to his queries, the Express-News reports, and it also does not explain how the FBI first learned of the threats.
Police say they linked Carrasco to the messages using his phone provider information, IP address, and Google account, according to the affidavit. The affidavit also states that Carrasco expressed a desire to obtain a firearm.
How the Tip Reached San Antonio
The FBI's National Threat Operations Section, which operates the bureau's tipline and evaluates potential threats, notified the Southwest Texas Fusion Center about the messages around August 28. The fusion center is an interagency intelligence-gathering hub managed by the San Antonio Police Department, per the affidavit as cited by the Express-News. Nationally, the FBI's tip-processing unit routes actionable reports from its National Threat Operations Center in West Virginia to local intelligence hubs and law enforcement agencies over secure administrative networks, according to the FBI.
Once a tip like this reaches San Antonio, it typically lands with the Southwest Texas Fusion Center's multi-agency Behavioral Threat Assessment Group, which meets three times weekly to evaluate potential mass-violence threats and coordinate responses across law enforcement, federal agents, and mental health specialists, the FBI has said. That same threat-assessment structure was credited last year with stopping a planned middle school shooting after a family tip led to the arrest of a 13-year-old student and his mother on terrorism charges, according to the San Antonio Report, marking Bexar County's first juvenile terrorism charge.
AI Chatbots Under Growing Scrutiny After Shootings
Carrasco's case adds to a widening national conversation about AI chatbots and school violence. In the April 2025 shooting at Florida State University, which killed two people and wounded several others, the alleged gunman reportedly used ChatGPT to discuss shooting plans. Florida Attorney General James Uthmeier opened a criminal investigation into OpenAI in April 2026 after prosecutors reviewed chat logs showing the FSU gunman used the chatbot for tactical advice on firearms and campus locations, according to the Florida Attorney General's office.
In May 2026, the family of a victim killed in that shooting filed a federal lawsuit against OpenAI, alleging the company failed to build safety features that would detect or report a user planning an attack despite months of explicit firearm-related prompts, as reported by The Guardian. That suit was filed by the widow of victim Tiru Chabba. Separately, the family of a 12-year-old injured in a Canadian school shooting that killed eight people sued OpenAI in March, according to the BBC.
Anthropic's own transparency policies, updated in July 2026, allow the company to disclose user account information to law enforcement without a subpoena or warrant if it believes there is an imminent risk of physical harm or death, according to Anthropic. The company has said it banned 11.4 million user accounts between January and June 2026 for violating usage policies that prohibit promoting or facilitating violence. It remains unclear whether Anthropic's safety systems flagged Carrasco's queries under that emergency exception or whether a third party reported him, a detail the arrest affidavit does not address.
Part of a Statewide Pattern
Carrasco's arrest lands amid a string of similar cases across Texas as the school year opened. Earlier this month, FBI Houston flagged an online threat that led Pasadena ISD police to arrest three high school students on third-degree felony terroristic threat charges, as Hoodline previously reported. Days later, federal intelligence helped track a Victoria teenager to a bus stop over a middle school threat post, and Willis High School in Montgomery County saw its own gun and shooting threats within a similar stretch of the semester.
Under Texas Penal Code § 22.07, making a terroristic threat that places the public in fear of serious bodily injury or interrupts public operations is a third-degree felony, according to legal analysis from Saputo Toufexis. A conviction carries between two and 10 years in state prison and a fine of up to $10,000, which is the range of criminal exposure Carrasco now faces in Bexar County.
