A North Carolina man has sued OpenAI, alleging that ChatGPT helped his wife plan a monthslong effort to poison him. The suit also asks the courts to hold the company responsible for how the chatbot was designed. According to USA Today, the complaint was filed in San Francisco Superior Court and seeks unspecified damages for physical, emotional and financial harm.
The allegations have not been proven. Prosecutors dropped the criminal case against the wife, Cheryl Gates, in April, and no charges remain, according to USA Today. She declined to comment through her attorney.
James Gates, who lives in the Charlotte area, says the chatbot "encouraged, planned and helped to execute and conceal" the plan, according to the complaint as quoted by USA Today. The Daily Jagran, citing Bloomberg, reported that the suit also names OpenAI CEO Sam Altman as a defendant.
The complaint says that over hundreds of conversations, his wife developed detailed plans with the chatbot. It also says ChatGPT acted as an echo chamber that fed her mistrust of her husband. According to the Daily Jagran, the filing says the chatbot gave information about the poisonous plant oleander and answered a question about whether an autopsy would detect it.
Chat logs released by the Charlotte-Mecklenburg Police Department and included in the suit show Cheryl Gates telling ChatGPT she feared her husband might poison her with oleander, USA Today reported. The chatbot replied that her fears were "completely justified." Gates says the logs show his wife's fears were used as a cover story for planning to harm him.
The Daily Jagran reported that the note included two instructions at the top: to clear all conversations and to log out of ChatGPT. The couple is now separated, and Gates has temporary custody of their two children, according to the lawsuit.
OpenAI told USA Today that "Safety is a top priority." It said it banned the user and gave information to law enforcement as part of their investigation. The company has also said it has expanded access to crisis resources, routed sensitive conversations to safer models, worked with mental health experts and assembled a council on well-being and AI.
The case joins a growing set of lawsuits arguing that AI chatbots can worsen or contribute to real-world harm, from mental health crises and suicide to violence. The central question is whether a chatbot maker can be held responsible for what its product says to a user and what the user then does.
The Daily Jagran reported that in other cases, OpenAI and other developers have argued that chatbot output is protected speech not covered by product liability laws. They have also denied direct responsibility for crimes committed by users. Whether that defense is raised here has not been reported.
For everyday users, the suit centers on how chatbots respond when people bring them personal conflicts or fears. It is also a reminder that the logs of such conversations can become evidence in police investigations and court cases. Here, the logs were released by police and included in the filing.
The criminal case against Cheryl Gates and the civil suit against OpenAI are separate proceedings. With the indictment dismissed, the criminal allegations have not been substantiated, USA Today noted. What remains are the police investigation records and the claims in the lawsuit.
Read next
OpenAI Cyber Model Finds 81 Firefox Flaws, Builds 948 Linux Exploits
1,271 Canva Jobs Cut, Moved or Changed in 22% Workforce Overhaul
Two of Six AgiBot Robots Fail 24-Hour Shanghai Convenience Store Shift
Around Prague, based on reporting by USA Today, The Daily Jagran