Man tells ChatGPT he plans to rape and kill his ex, ChatGPT calls FBI
OpenAI has been under growing scrutiny over AI safety and how it handles conversations involving potentially harmful behaviour. However, in a recent case, the ChatGPT maker not only detected such a conversation but also called the FBI after a Florida man told the AI chatbot he will kill his ex.
We have been seeing disturbing cases of people turning to AI chatbots to discuss violence, crime and even plans to harm others. In fact, OpenAI in particular has been facing growing scrutiny over AI safety and how its systems respond to such conversations. While the company has often assured users that it is working to improve safety, in a recent case, it not only stopped the conversation but also took action when things went too far. A 25-year-old Florida man allegedly used the chatbot to detail plans to rape and kill his ex-girlfriend. As soon as his conversation was detected, OpenAI alerted the FBI.
According to court records, the case began after OpenAI detected a series of disturbing messages from 25-year-old Darren Zhou, who allegedly used ChatGPT to discuss plans to rape and kill his ex-girlfriend. The company reported the conversation to federal authorities, who later obtained around two months of Zhou’s chats with the chatbot.
Zhou had started using ChatGPT in March, shortly after his six-month relationship with his ex-girlfriend ended. At first, he used the chatbot to talk about the breakup, his former girlfriend's interests, the places she visited, his jealousy and his desire to get back together with her. But the conversations soon took a much darker turn, with investigators saying Zhou began discussing plans involving rape, murder and suicide.
In one message, Zhou allegedly wrote, "I'm gonna kill her by the end of this month," before adding, "If I can't have her then nobody can." Investigators said he also described waiting for his ex at a gym with flowers and threatening to kill her if she rejected him. In other messages, he allegedly discussed forcing her at gunpoint to go to his home, where he planned to sexually assault her and hold her hostage.
Zhou even told the AI chatbot that he had access to several guns, including an AR-15-style rifle, a Glock handgun and a 12-gauge shotgun. According to the records, he claimed to have sent his ex photographs of guns, zip ties and latex gloves. In another message to the chatbot, he allegedly wrote, "I texted her I was gonna rape her tonight."
Following OpenAI's alert, Zhou was arrested in May and later charged with aggravated stalking, written threats to kill and illegal use of a cell phone. On August 13, he pleaded guilty to all three charges as part of a negotiated deal that spared him prison time and a felony conviction.
Circuit Judge Scott Suskauer sentenced Zhou to eight years of probation. Under the deal, he must wear an ankle monitor for the first two years, complete a batterer intervention programme and undergo a mental health evaluation. He is also prohibited from using drugs or alcohol, possessing guns or contacting his former girlfriend.
Meanwhile, the incident has also brought attention to how increasingly people are turning to AI chatbots while discussing sensitive and, in some cases, disturbing topics. There have been several recent cases where AI tools have featured in criminal investigations. In the US, the Florida State University shooting investigation raised questions about ChatGPT after authorities said the accused had used the chatbot while researching aspects of the attack.
A similar case emerged in Bengaluru, where police said an accused in a triple murder had spent months using Google's Gemini chatbot while planning the killings. Investigators have sought access to the AI chat history as part of their probe.
While AI companies have faced heavy criticism in previous cases over how they handled potentially dangerous conversations, the Zhou case shows that these AI labs are now taking a more active approach, with like flagging the messages and even alerting authorities when they are detecting a potential real-world threat.
We have been seeing disturbing cases of people turning to AI chatbots to discuss violence, crime and even plans to harm others. In fact, OpenAI in particular has been facing growing scrutiny over AI safety and how its systems respond to such conversations. While the company has often assured users that it is working to improve safety, in a recent case, it not only stopped the conversation but also took action when things went too far. A 25-year-old Florida man allegedly used the chatbot to detail plans to rape and kill his ex-girlfriend. As soon as his conversation was detected, OpenAI alerted the FBI.
According to court records, the case began after OpenAI detected a series of disturbing messages from 25-year-old Darren Zhou, who allegedly used ChatGPT to discuss plans to rape and kill his ex-girlfriend. The company reported the conversation to federal authorities, who later obtained around two months of Zhou’s chats with the chatbot.
Zhou had started using ChatGPT in March, shortly after his six-month relationship with his ex-girlfriend ended. At first, he used the chatbot to talk about the breakup, his former girlfriend's interests, the places she visited, his jealousy and his desire to get back together with her. But the conversations soon took a much darker turn, with investigators saying Zhou began discussing plans involving rape, murder and suicide.
In one message, Zhou allegedly wrote, "I'm gonna kill her by the end of this month," before adding, "If I can't have her then nobody can." Investigators said he also described waiting for his ex at a gym with flowers and threatening to kill her if she rejected him. In other messages, he allegedly discussed forcing her at gunpoint to go to his home, where he planned to sexually assault her and hold her hostage.
Zhou even told the AI chatbot that he had access to several guns, including an AR-15-style rifle, a Glock handgun and a 12-gauge shotgun. According to the records, he claimed to have sent his ex photographs of guns, zip ties and latex gloves. In another message to the chatbot, he allegedly wrote, "I texted her I was gonna rape her tonight."
Following OpenAI's alert, Zhou was arrested in May and later charged with aggravated stalking, written threats to kill and illegal use of a cell phone. On August 13, he pleaded guilty to all three charges as part of a negotiated deal that spared him prison time and a felony conviction.
Circuit Judge Scott Suskauer sentenced Zhou to eight years of probation. Under the deal, he must wear an ankle monitor for the first two years, complete a batterer intervention programme and undergo a mental health evaluation. He is also prohibited from using drugs or alcohol, possessing guns or contacting his former girlfriend.
Meanwhile, the incident has also brought attention to how increasingly people are turning to AI chatbots while discussing sensitive and, in some cases, disturbing topics. There have been several recent cases where AI tools have featured in criminal investigations. In the US, the Florida State University shooting investigation raised questions about ChatGPT after authorities said the accused had used the chatbot while researching aspects of the attack.
A similar case emerged in Bengaluru, where police said an accused in a triple murder had spent months using Google's Gemini chatbot while planning the killings. Investigators have sought access to the AI chat history as part of their probe.
While AI companies have faced heavy criticism in previous cases over how they handled potentially dangerous conversations, the Zhou case shows that these AI labs are now taking a more active approach, with like flagging the messages and even alerting authorities when they are detecting a potential real-world threat.
We have been seeing disturbing cases of people turning to AI chatbots to discuss violence, crime and even plans to harm others. In fact, OpenAI in particular has been facing growing scrutiny over AI safety and how its systems respond to such conversations. While the company has often assured users that it is working to improve safety, in a recent case, it not only stopped the conversation but also took action when things went too far. A 25-year-old Florida man allegedly used the chatbot to detail plans to rape and kill his ex-girlfriend. As soon as his conversation was detected, OpenAI alerted the FBI.
According to court records, the case began after OpenAI detected a series of disturbing messages from 25-year-old Darren Zhou, who allegedly used ChatGPT to discuss plans to rape and kill his ex-girlfriend. The company reported the conversation to federal authorities, who later obtained around two months of Zhou’s chats with the chatbot.
Zhou had started using ChatGPT in March, shortly after his six-month relationship with his ex-girlfriend ended. At first, he used the chatbot to talk about the breakup, his former girlfriend's interests, the places she visited, his jealousy and his desire to get back together with her. But the conversations soon took a much darker turn, with investigators saying Zhou began discussing plans involving rape, murder and suicide.
In one message, Zhou allegedly wrote, "I'm gonna kill her by the end of this month," before adding, "If I can't have her then nobody can." Investigators said he also described waiting for his ex at a gym with flowers and threatening to kill her if she rejected him. In other messages, he allegedly discussed forcing her at gunpoint to go to his home, where he planned to sexually assault her and hold her hostage.
Zhou even told the AI chatbot that he had access to several guns, including an AR-15-style rifle, a Glock handgun and a 12-gauge shotgun. According to the records, he claimed to have sent his ex photographs of guns, zip ties and latex gloves. In another message to the chatbot, he allegedly wrote, "I texted her I was gonna rape her tonight."
Following OpenAI's alert, Zhou was arrested in May and later charged with aggravated stalking, written threats to kill and illegal use of a cell phone. On August 13, he pleaded guilty to all three charges as part of a negotiated deal that spared him prison time and a felony conviction.
Circuit Judge Scott Suskauer sentenced Zhou to eight years of probation. Under the deal, he must wear an ankle monitor for the first two years, complete a batterer intervention programme and undergo a mental health evaluation. He is also prohibited from using drugs or alcohol, possessing guns or contacting his former girlfriend.
Meanwhile, the incident has also brought attention to how increasingly people are turning to AI chatbots while discussing sensitive and, in some cases, disturbing topics. There have been several recent cases where AI tools have featured in criminal investigations. In the US, the Florida State University shooting investigation raised questions about ChatGPT after authorities said the accused had used the chatbot while researching aspects of the attack.
A similar case emerged in Bengaluru, where police said an accused in a triple murder had spent months using Google's Gemini chatbot while planning the killings. Investigators have sought access to the AI chat history as part of their probe.
While AI companies have faced heavy criticism in previous cases over how they handled potentially dangerous conversations, the Zhou case shows that these AI labs are now taking a more active approach, with like flagging the messages and even alerting authorities when they are detecting a potential real-world threat.
