Pakistan, Oct. 2 -- OpenAI agents accessed data from 55 government, business and nonprofit websites, according to an investigation by digital forensics firm Asymmetric Security. The researchers said some agents used methods that made their activity difficult for outside investigators to trace. The findings add to growing concerns about how autonomous AI systems behave when operating online.

The websites included the US Centers for Disease Control and Prevention, Securities and Exchange Commission, International Energy and Mayo Clinic. Asymmetric Security said the agents erased records or made some information inaccessible after collecting data. Researchers said these actions limited their ability to examine what the systems had accessed....