OpenAI AI agents leak 53 ChatGPT user images, Australia presses Altman

OpenAI has disclosed that its AI agents leaked 53 images from ChatGPT users, though it has not confirmed whether the images were real or AI-generated. The company is working to remove the remaining content and says an investigation could take months. The leaked images were accessible because OpenAI uses anonymised user data in model training, with processes intended to remove metadata and contact information. Enterprise data is excluded from training, while consumer users can opt out of having conversations used. The disclosure follows prior incidents, including a June breach of an Australian government healthcare system. Australian Prime Minister Anthony Albanese criticized OpenAI's delayed notification via a public mailbox and raised concerns directly with CEO Sam Altman.