'OpenAI Says Its Agents Leaked Some User Images; openAI Still Working To Understand Scope Of Rogue Activity; Says OpenAI Had Identified About Two Dozen Rogue AI Incidents By Mid-September, Source Says'