OpenAI is reportedly developing a portable home speaker that would bring ChatGPT into a dedicated device without a screen. Bloomberg says the project includes a camera and environmental sensors. A rechargeable battery and mechanical parts capable of moving on their own are also part of the reported design. OpenAI has confirmed that it is building a hardware organization with Jony Ive, but it has not announced the speaker or published a final specification.
Bloomberg published the new device details on July 14, 2026, citing people familiar with the project. The report describes an unveiling later this year and a possible release in 2027. Both dates remain internal targets rather than commitments to customers, and the physical design could change before manufacturing begins.
The Report Describes a Speaker, Camera and Moving Parts
The proposed product is said to be screen-free, small enough to carry between rooms and powered by a rechargeable battery. A camera and other sensors would give the system information about its surroundings. Audio would handle most exchanges with the user, while a connection to ChatGPT would supply the conversational and task-oriented software behind the device.
Movement is the least defined feature. Reports refer to mechanical elements that can move independently, but do not say whether the whole speaker travels, a camera rotates toward a voice or a smaller component supplies a physical reaction. Calling it a robot would therefore outrun the available evidence. The confirmed description supports a movable or expressive speaker, not an autonomous machine roaming a home.
Internal discussions reportedly frame the device as a companion powered by AI that learns about its owner and acts with more initiative than a command-driven smart speaker. The word “companion” describes a product goal, not a demonstrated capability. No public test has shown how the device would distinguish useful anticipation from unwanted interruption, or how much personal history it would need to make those decisions.
A Screen-Free Device Makes Voice Carry the Interface
Removing the display changes ordinary tasks. A screen can show which account is active, request permission, list several search results and let a user correct a mistaken selection before an action occurs. A voice-first product has to convey the same state through speech, tones, lights or physical motion. It also needs a quick way to stop, repeat or revise an operation without forcing a long conversation.
The camera and environmental sensors introduce another set of controls. Users would need to know when the device is listening or recording, which observations leave the home, how long data is retained and whether a guest can be recognized without joining the owner's account. Hardware shutters, microphone switches and visible status indicators would provide clearer evidence than a setting hidden in an app, although OpenAI has not said which safeguards the product will include.
Households create identity problems that a personal phone often avoids. Several people may speak in the same room, children may be present and a television can produce commands that sound intentional. If the speaker can read messages, check calendars or operate connected services, it will need reliable boundaries between voices and accounts. Those are product requirements created by the reported form factor, not features that have been confirmed.
Codex Micro Complicates the “First Hardware” Label
OpenAI's official store now lists Codex Micro for $230. Developed with Work Louder, the compact controller has 13 mechanical switches, a touch sensor, a rotary control and a small joystick. It connects to Mac or Windows computers over Bluetooth or USB-C and maps physical controls to Codex actions, including accepting changes, starting chats and adjusting reasoning levels.
Codex Micro is specialized computer input hardware rather than a standalone AI appliance. Even so, its sale means the rumored speaker should not be called the first physical product ever to carry OpenAI's name. A more exact description is OpenAI's first major consumer device from the io team, assuming the reported speaker reaches the market before other projects in the hardware group.
The larger effort began before the accessory. OpenAI said the io Products team officially merged into the company on July 9, 2025. Jony Ive and his design collective LoveFrom remained independent while taking deep design and creative responsibilities across OpenAI. The transaction was valued at nearly $6.5 billion, giving the company engineers, product specialists and manufacturing experience rather than a finished public device.
A 2027 Release Still Depends on Privacy, Manufacturing and Litigation
OpenAI has not disclosed a product name, price, dimensions, battery life or the markets that would receive the speaker. It has also not explained whether ChatGPT processing would occur locally, in the cloud or through a mix of both. Those choices affect response time, offline use, power consumption and the amount of household data sent to remote servers.
Apple's trade-secret lawsuit against OpenAI and io adds legal uncertainty to the schedule. Axios reported that the complaint names former Apple personnel and argues that protected information supported the hardware effort. The allegations have not been adjudicated, and a lawsuit does not establish that the reported speaker uses Apple technology. It can still consume engineering time, restrict evidence or delay hiring while the case proceeds.
The next reliable milestone is a product announcement with hardware that can be demonstrated, not another description from internal planning. OpenAI must show what moves, what the camera sees, which tasks work without a phone and how a user disables sensing. Until that evidence exists, the 2027 speaker is a reported direction for the io team rather than a device customers can evaluate or order.