OPENAI is reportedly preparing to unveil a persistent AI assistant codenamed “O” at its DevDay conference in San Francisco on 29 September. The report is based on confidential sources and an examination of ChatGPT configuration files, which allegedly found display settings using the name “O” and email configurations ending in “-o”. These findings suggest a system designed to operate in the background rather than only responding to user prompts within a chat interface.
According to the report, “O” could have its own system identity and email address, allowing it to respond to external triggers and carry out tasks autonomously. The project is said to be connected to “Project Aeon”, described as an effort involving long-term memory and scheduled task execution, although the assistant’s initial permissions and capabilities remain unclear.
The article also links the project to unconfirmed reports that OpenAI is working with former Apple design chief Jony Ive on a physical AI device, potentially with a circular or “donut” design. No hardware launch is confirmed.
The claims point to a broader industry shift towards ambient AI agents that can monitor information, manage schedules or perform other tasks without continuous user interaction. However, OpenAI’s alleged plans remain unconfirmed, and the report does not provide evidence of exploitation, a security incident or an available product.