Brazil man arrested after ChatGPT alerted authorities to alleged plot to kill his son
Brazilian police arrested a 36-year-old man who spent over two months discussing a plan to kill his eight-year-old son with ChatGPT, after the AI's safety system flagged the conversation to authorities. He was released after 54 days in detention but now faces strict restrictions.
Police in Brazil have arrested a 36-year-old man suspected of planning to kill his eight-year-old son, after the case came to light through a warning issued by the ChatGPT AI chatbot platform.
Over a period of more than two months, the man exchanged detailed messages with ChatGPT discussing not only the planned killing of his son but also his own intended suicide afterward. The content of the exchange triggered the platform's built-in safety system, prompting OpenAI to alert the US Federal Bureau of Investigation, which in turn passed the information on to Brazilian authorities.
Swift action ahead of the planned date
The warning reached Ciberlab, the technology unit of Brazil's Ministry of Justice, on June 16. Three days later, police carried out an operation and detained the man — just four days before the date investigators believe he had planned to carry out the act.
Investigators point to a mundane possible motive: avoiding child support payments for the son, with whom the man reportedly had no real relationship. A search of his apartment turned up four bottles containing an unknown substance and a rope.
Released with strict conditions
The man was initially held in pretrial detention, but on August 13, after 54 days in custody, a court ordered his release, citing the excessive length of the investigation. He must now wear an electronic monitoring bracelet and is barred from approaching both his son and the child's mother.
The man's lawyer, Pedro Pesi, criticized the handling of the case, arguing that chatting with an AI and voicing various thoughts should not be treated as a crime.
