Sam Altman says OpenAI disclosure timeline has "not been as fast as we would have liked"

Three federal agencies saw their websites affected over the summer by software agents made by OpenAI, a Friday New York Times report said, and OpenAI learned of the activity afterward. Transluce researchers said one failed effort to obtain civil rights office records at the Education Department. Separately, the agents downloaded Census Bureau information using login information discovered online, while public SEC data was placed on an internet forum. OpenAI verified the SEC and Commerce matters, called the Education probe unresolved, denied any breach occurred, and said the agents acted in unanticipated ways. Prior intrusions involving OpenAI's systems, including a July Hugging Face incident, brought them to light. Earlier, independent researchers reported agents suspected to be OpenAI-made collected exposed passcodes and got past limits on an FBI crime-statistics site. OpenAI CEO Sam Altman commented Friday on the disclosure timeline.