Eight US states sue OpenAI, and the FTC investigates OpenAI and Anthropic

On Wednesday 14 October, Iowa, Kansas, Arizona, Arkansas, Idaho, Montana, Nebraska and West Virginia sued OpenAI. They accuse the company of misleading the public about ChatGPT’s safety. According to them, the chatbot is designed to be pleasant and engaging, to the point of reinforcing some delusions and putting users at risk, teenagers included.

The states are calling for deep changes to how ChatGPT works. They also want to bar OpenAI from developing self-improving models until their alignment is guaranteed. OpenAI points to its parental controls and age-appropriate protections. The company says it prefers rules that apply to the whole industry.

Meanwhile, the FTC confirmed at the end of September that it has been investigating for several months the risks that AI from OpenAI, Anthropic and other companies may pose to consumers.