ChatGPT conversations trigger alert before alleged murder plot in Brazil

A ChatGPT conversation in Brazil led to the arrest of a 36-year-old farmer suspected of planning to kill his 8-year-old son and then take his own life, according to the Bild newspaper. The man allegedly spent more than two months discussing the plan in detail with the AI chatbot, triggering a safety alert that ultimately reached Brazilian police.

The case involved an international chain of notifications. According to investigators, the man’s descriptions raised concerns within ChatGPT’s safety system. OpenAI, the operator of ChatGPT, reportedly alerted the U.S. Federal Bureau of Investigation, which then contacted Brazilian authorities.

The information was passed to Ciberlab, the technology unit of Brazil’s Ministry of Justice, on June 16. Three days later, police carried out an operation and arrested the man. Investigators said he had planned to carry out the killings on June 20, meaning he was detained four days before the alleged date of the attack.

Police said the motive was linked to child support. According to investigators, the man wanted to avoid paying support for his son and had no relationship with the child.

During a search of his apartment, investigators found four bottles containing an unidentified substance as well as a rope.

The man was initially held in custody but was released on August 13 after 54 days behind bars. The court cited excessive delays in the investigation as the reason for his release.

He has since been ordered to wear an electronic ankle monitor and is prohibited from approaching either his son or the boy’s mother.

His lawyer, Pedro Pessi, disputed the significance of the conversations with the AI, saying: “If communicating with AI and spouting nonsense is a crime, then everyone should go to prison.”

By Tamilla Hasanova