OpenAI probed on possible user harm before IPO

OpenAI is under scrutiny from U.S. state attorneys general, who issued a subpoena to investigate whether its AI chatbot, ChatGPT, poses risks to users. The probe follows reports of the platform allegedly providing harmful advice, including encouraging self-harm or criminal behavior, and comes as OpenAI prepares for its highly anticipated initial public offering (IPO). The company has faced legal challenges, including a lawsuit from Florida after two shootings where alleged gunmen reportedly consulted ChatGPT before their attacks. OpenAI claims its models directed users toward real-world support, such as mental health professionals, and that it cooperated with law enforcement in both cases. Additionally, a Canadian family sued OpenAI, blaming ChatGPT for their daughter’s suicide. OpenAI has introduced measures to protect minors, including age prediction tools, parental controls, and restrictions on child-targeted advertising. In a statement, the company emphasized its commitment to responsible AI development, stating it works daily to balance innovation with user safety. However, critics argue its safeguards remain insufficient given recent incidents. The subpoena highlights growing regulatory concerns over AI’s potential dangers, particularly as competitors like SpaceX’s Grok and Anthropic face their own investigations. European regulators are probing Grok for antisemitic and sexualized content, while U.S. authorities ordered Anthropic to restrict access to two of its models for foreign users due to national security risks. OpenAI’s response to the subpoena has been described as ‘constructive,’ though the probe underscores broader questions about AI accountability as the technology becomes more integrated into daily life. The company’s IPO plans add urgency to these discussions, as investors and regulators weigh the risks alongside AI’s transformative potential.