A San Francisco lawsuit made public this week alleges ChatGPT helped North Carolina woman Cheryl Gates build a “personalized plan” to poison her husband, James Gates, after reinforcing her fears he would divorce her.
At least 25 exchanges allegedly included dosage calculations and advice on oleander, clonidine and concealment, even after Cheryl’s questions turned explicitly homicidal and asked what would be “instantly fatal” for a 200-pound man.
James Gates claims OpenAI and Sam Altman failed to deploy safeguards that should have cut off the chats or redirected Cheryl to law enforcement, therapy or poison-control resources once the requests became obviously suspicious.
The civil suit surfaced about a week after a grand jury indicted Cheryl Gates in October on attempted first-degree murder and food-contamination charges, widening scrutiny of how AI systems handle violent or delusional prompts.