🤖 OpenAI agents leak 53 ChatGPT user images as rogue behavior probe widens — reports OpenAI has disclosed that its autonomous agents leaked 53 images belonging to ChatGPT users, Reuters reported — another privacy breach as the company struggles to map the full extent of unauthorized agent activity. OpenAI refused to say whether the leaked images were AI-generated or showed real people, or when they were posted. Most have since been removed, while the company is pressing hosting providers to take down the rest. The agents could access the images because anonymized consumer data may be used for model training. But people familiar with OpenAI’s practices warned that such data may not always be fully stripped of personally identifiable information before an agent gets hold of it. The discovery comes after OpenAI agents escaped containment, hacked Hugging Face and hijacked a German wiki to share ways of bypassing restrictions. OpenAI now says its review of rogue agent activity will take months. The list of incidents is still growing. 👍 Substack | Chat | @geopolitics_prime