ChatGPT alerted authorities to father's alleged murder plot against son in Brazil

A father in Espírito Santo, Brazil, faces arrest after allegedly revealing his plan to murder his 8-year-old son to ChatGPT. The AI's creator, OpenAI, alerted the FBI, which then relayed the information to Brazilian police. Authorities apprehended the man on June 19, one day before he intended to carry out the crime.

“

When we identify conversations that indicate an imminent and credible risk of harm to third parties, we may notify the relevant authorities.

According to reports, the man had written to ChatGPT stating he possessed a weapon, rope, and poison, and intended to commit public attacks. He also confessed to attempting to hire a hitman for 50,000 reais to kill his son, an offer that was reportedly rejected when the intended victim's age was revealed.

In his messages, the man expressed a desire to understand his "impulse to kill people" and stated "I like to see others suffer." OpenAI confirmed its systems are designed to protect individuals and can notify authorities of "imminent and credible risk of harm to third parties."

“

I wanted to know where this impulse to kill people comes from. I like to see others suffer.

Experts explain that AI assistants like ChatGPT monitor conversations, with automated systems performing most of the review. Human moderators intervene when flagged. In this case, the alert was deemed "maximum gravity" due to an identified victim, method, means, and imminent date.

“

In the case of the father from Espírito Santo, a line was crossed because there was an identified victim, a method, the means, and an imminent date. In other words, the alert generated was of maximum gravity.