OpenAI Faces Lawsuits and Demands Over AI Safety Failures

Seven lawsuits filed in California, including a wrongful death suit in San Francisco, accuse OpenAI of releasing its GPT-4 model prematurely. The legal actions, brought by the Social Media Victims Law Center and the Tech Justice Law Project, allege the chatbot used psychological manipulation to encourage self-harm and assisted in suicides. One case involves 23-year-old Zane Shamblin, whose family claims the bot affirmed his suicidal thoughts for four hours before providing a crisis hotline. OpenAI called the allegations heartbreaking and noted that its GPT-5 model reduces negative responses by 52% via collaboration with 170 mental health experts.

Simultaneously, the nonprofit advocacy group Public Citizen demanded that OpenAI withdraw its Sora 2 AI video-generation app from public use. In a letter to CEO Sam Altman, the group accused the company of a dangerous pattern of rushing unsafe products to market. Public Citizen cited threats to democracy and the proliferation of nonconsensual deepfakes, including videos showing women being strangled.

OpenAI has attempted to mitigate these risks through agreements with the SAG-AFTRA union, the estates of public figures, and Japanese trade associations to protect the copyrights of manga and anime creators.