First OpenAI hardware device sounds half-speaker, half-robot

OpenAI has been talking up its move into hardware for some time, but the shape of its first device has remained unclear. According to a new report, the company is starting with a movable, screenless smart speaker designed to act as a physical manifestation of ChatGPT in the home.

Bloomberg reports that the details come from people familiar with the project, who asked not to be identified because it hasn’t been announced. The device is still under development and is intended to serve as a humanlike AI companion, capable of controlling smart-home appliances, playing media, answering questions, responding to messages, and drawing on ChatGPT’s broader capabilities.

The speaker will reportedly include a camera and other sensors to understand its surroundings, and will also use personal information, such as emails, to learn more about its owner. OpenAI is said to want the device to become more personalized and proactive over time, including anticipating needs and surfacing information proactively. Mechanical elements may also be involved to make it feel more alive.

A rechargeable battery would let users carry the device from room to room — there’s no suggestion that the mechanical parts are some kind of wheels or legs. Bloomberg says it could be moved into the kitchen for cooking help, taken into another room while doing chores, and later used for music. Its voice features are expected to rely on GPT-Live, the more advanced version of ChatGPT Voice Mode that OpenAI introduced this month.

OpenAI is reportedly aiming to unveil the device this year before releasing it in 2027, although those plans could still change. Apple is seeking an injunction against OpenAI’s hardware business as part of a lawsuit alleging trade-secret theft — a move that could affect the timeline. The company is said to have around five hardware products in development, including wearables, home robotics, and a possible smartphone replacement.