OpenAI Faces Multistate U.S. Probe as IPO Nears

OpenAI is under mounting pressure in the United States as it moves toward an initial public offering. A coalition of state attorneys general has opened a broad investigation into the company, focusing on ChatGPT’s child safety protections, user engagement tactics, consumer and health data handling, and how the system interacts with users.

According to The Wall Street Journal, OpenAI received a subpoena on Friday from the New York attorney general’s office requiring documents and information on a wide range of issues. Investigators are also examining the company’s advertising practices, the ways it keeps users engaged and encourages continued use, and its policies toward minors and older adults. One unusual area of interest is the model’s tendency toward sycophancy, or pleasing users rather than giving accurate answers, a behavior researchers say can reinforce dangerous thinking and bad decisions.

OpenAI did not say which other states are involved or what additional materials it must provide. The company said it takes the prosecutors’ concerns seriously and will cooperate. It added that AI is a powerful new technology and that it is working every day to deliver its benefits safely and responsibly. OpenAI also said ChatGPT now includes protections for minors and people in distress, including referrals to human help, a system to estimate users’ ages, parental tools, and a ban on ads aimed at children.

The probe adds to several legal and regulatory challenges already facing the company. Earlier this month, Florida’s attorney general sued OpenAI and CEO Sam Altman, accusing them of ignoring internal and external safety warnings and exposing children to a dangerous product. OpenAI is also fighting copyright and harm claims, including suits tied to alleged suicides involving ChatGPT users. The company recently won a high-profile case brought by cofounder Elon Musk, though his lawyer plans to appeal. Altman also apologized to residents of Tumbler Ridge, British Columbia, after OpenAI reportedly detected unusual activity in a suspect’s ChatGPT account in a February mass shooting but did not alert law enforcement. The investigation comes just days after OpenAI said it had confidentially filed IPO documents.