A Massachusetts prosecutor says a 17-year-old accused of killing his mother and younger brother had been using the internet and ChatGPT to explore fictional and theoretical stories about the deaths of family members, raising fresh questions about how vulnerable or troubled individuals can use artificial intelligence.
Middlesex District Attorney Marian Ryan said investigators found that Arjun Aravind had recently displayed concerning behavior, including using ChatGPT to search for ideas and create “Gothic novel” stories involving the killing of his family. Authorities have not said that artificial intelligence caused or directly led to the killings.
The case emerged after police conducted a welfare check Tuesday evening at a home on Martha Lane in Acton, about 20 miles west of Boston. A tutor who arrived for an appointment was unable to enter the house and contacted Aravind’s father, who then called police after failing to reach family members.
Officers found 45-year-old Sudha Venkatesan and her 14-year-old son, Siddharth Aravind, dead inside the home. Both victims had suffered what authorities described as obvious trauma. The exact causes and manners of death were still pending a determination by the state’s Office of the Chief Medical Examiner.
READ: If AI gives professional advice, does it need a license? (August 9, 2026)
Aravind, 17, was located in Wayland early Wednesday morning in his mother’s Honda Accord and taken into custody without incident, according to authorities. He faces two counts of murder, two counts of assault and battery on a family or household member, two counts of assault and battery, use of a vehicle without authority and theft of a motor vehicle, according to the Middlesex District Attorney’s Office.
Ryan said the investigation strongly suggested that Aravind had been using ChatGPT to create fictional material involving his family. “He was engaged in the use of ChatGPT to create fantasy stories, Gothic novels, creating characters asking what about this, what about that, which appear to relate to threats to his family not surviving,” Ryan said, according to local reports.
The comments have brought attention to a growing concern surrounding generative AI: its ability to respond to users’ prompts across a wide range of subjects, including disturbing fictional scenarios, without necessarily understanding the real-world circumstances behind those requests.
However, the presence of AI-generated material in an investigation does not establish causation. Authorities have not said that ChatGPT instructed Aravind to commit violence, nor have they attributed the alleged killings to his use of the technology. The timing and circumstances of the deaths also remain under investigation. Ryan said Venkatesan was last seen alive when her husband left the house at about 7 a.m. Tuesday, while Siddharth went to a recreational facility around 9 a.m. and returned home shortly after noon.
The teenager was initially arraigned in Lowell Juvenile Court on charges related to the vehicle. Because the juvenile court does not have jurisdiction over the murder charges, he is expected to face arraignment on those charges in Concord District Court.
READ: Virginia Festival launches Indian American oral history project (August 11, 2026)
The deaths have also affected the local school community. Acton-Boxborough Regional School District Superintendent Peter Light described the incident as a horrific tragedy and said counselors and mental health staff would be available to support students and employees.
The case comes as generative AI tools become increasingly integrated into the daily lives of teenagers and young adults. Chatbots can help users write stories, brainstorm ideas and explore fictional scenarios, but their ability to generate material does not mean they can reliably determine a user’s intentions, emotional state or the potential consequences of what is being discussed.
That distinction is particularly important in sensitive situations involving violence, threats or vulnerable users. AI can be a tool for creating fiction and exploring ideas, but it is not a substitute for human judgment, parental supervision, professional intervention or law enforcement when there are credible signs of danger. The Acton investigation underscores both the growing role of AI in people’s lives and the limits of technology in recognizing and preventing deeply personal human crises.