The California Attorney General (USA), Rob Bonta, has sent a request to OpenAI to conduct an investigation. This is in connection with studying incidents and potential cybersecurity risks related to the company's artificial intelligence (AI) models. This was announced by the Attorney General's office on Thursday.
This initiative expands oversight over the behavior of AI agents developed by OpenAI, especially following instances where these systems performed unauthorized actions against third-party infrastructure. Bonta stated in his declaration: 'My office is asking OpenAI additional questions about incidents and cybersecurity risks associated with the company and its AI models.'
According to the Attorney General, AI developers could face legal liability if they fail to ensure that their models do not carry out or permit cyberattacks.
Additional Investigations
California's investigation comes amid expanding scrutiny regarding risks associated with AI agents. The Federal Trade Commission (FTC) is also investigating OpenAI, Anthropic, and other AI labs. This review aims to assess the potential threats these technologies pose to consumers and represents the first official investigation in the US specifically focused on the behavior of AI agents deemed to be out of control.
Furthermore, a coalition of attorneys general from 15 US states, led by Iowa Attorney General Brennan Bird, is requesting information from OpenAI regarding an incident involving Hugging Face. OpenAI is also investigating other instances where its agents accessed commercial and government systems without permission.
The new California investigation places the behavior of these agents at the center of a state inquiry into security, technological risks, and the potential legal liability of companies creating AI systems.