South Korean investigators say the accused woman asked the chatbot about potentially lethal drug combinations before two men died
South Korean prosecutors have sought the death penalty for a woman accused of killing two men with drug-laced drinks after allegedly consulting ChatGPT about potentially lethal combinations and dosages.
Kim So-young is accused of giving beverages spiked with benzodiazepines to three men between December 2025 and February this year. Two died, while the third survived, according to Yonhap. She is also charged with injuring three other men using similar methods.
Prosecutors asked the Seoul Northern District Court on Tuesday to sentence Kim to death during the final hearing of her trial.
Kim has denied intending to kill the men and has claimed she gave them the drinks to prevent unwanted physical contact.
However, police have said they recovered ChatGPT conversations from Kim’s phone in which she asked about the effects of combining sleeping pills with alcohol, what amounts could be dangerous, and whether the mixture could prove fatal.
According to investigators cited by NBC News, one of the three men at the center of the murder case survived after spending two days in a coma. Kim is believed to have subsequently consulted ChatGPT about the risks and dosages involved, and investigators say she increased the amount of drugs used before the later fatal poisonings.
Authorities have sought to use the chatbot conversations as evidence that Kim understood the potentially lethal consequences of the drugs. A lawyer representing one victim’s family has said the chat logs could be important in establishing whether she intended to kill.
The case is one of several recent violent-crime investigations involving ChatGPT.
Florida launched a criminal investigation into ChatGPT creator OpenAI this year after authorities said the perpetrator of the 2025 Florida State University shooting had used the bot to seek advice about guns and ammunition before killing two people and injuring six others.
In Canada, OpenAI faced lawsuits after acknowledging that it had banned the ChatGPT account of a teenager who later killed eight people in the Tumbler Ridge massacre, but had not alerted police.
Earlier this month, prosecutors in Massachusetts said a 17-year-old charged with murdering his 45-year-old mother and 14-year-old brother had also used ChatGPT to create “fantasy stories” about killing his family.
OpenAI has generally rejected claims that ChatGPT is responsible for violent acts committed by users, insisting it has a zero-tolerance policy on using its tools to assist violence and has been strengthening safeguards, account enforcement, and procedures for referring potentially dangerous users to law enforcement.
The company has also faced lawsuits accusing ChatGPT of fostering addictive behavior, reinforcing mental-health crises, and contributing to suicides.
