OpenAI’s First AI Device: Confirmed Facts and Rumours

OpenAI has confirmed a consumer-hardware project, not a product buyers can order. It has announced no device name, specifications, price or retail date. Bloomberg’s most detailed report points to a portable, screen-free speaker with Jony Ive’s team. Other reporting mentions glasses, a lamp, earbuds and even a phone, but these ideas do not form an official product roadmap.

What OpenAI has actually confirmed

OpenAI’s 2025 announcement, updated in July, says the io Products team merged with the company, while Ive and LoveFrom remained independent and assumed design and creative responsibilities across OpenAI. The announcement said early explorations had become tangible designs but named no device. In August 2026, chief financial officer Sarah Friar described AI-native devices as part of a wider system that includes models, software and computing infrastructure. That shows strategic intent, not a product specification.

Executives have pointed to a possible reveal, not promised a retail launch. In January, chief global affairs officer Chris Lehane told Axios OpenAI was on track to unveil its first device in the second half of 2026, but he would not commit to it going on sale that year. In April, Friar told The Associated Press consumer hardware was expected toward year-end. At DevDay on September 29, Altman teased hardware as something worth waiting for, Axios reported, but presented no device. OpenAI’s official recap listed more than 20 announcements without hardware specifications.

The speaker is the clearest reported lead

Bloomberg reported in July, citing people familiar with the project, that OpenAI’s first consumer device is a mobile, screen-free speaker intended as an AI companion at home. Sources described possible ChatGPT answers, smart-home control, media playback and message responses. The device was still under development. Earlier reporting summarized by MacRumors added speculative camera, facial-recognition and purchasing capabilities, while Bloomberg later described a camera and sensors, moving elements and a battery-powered form. None has been confirmed by OpenAI.

A possible hardware family, but no roadmap

The Information reported in February that a hardware team of more than 200 was exploring a speaker, smart lamp and possible glasses; these concepts were early and some could be cancelled. Bloomberg later said roughly five products were in development, with the speaker first. Reported prices also conflict: The Information put the speaker at $200–$300, while Bloomberg’s August update put it at $300–$400. The gap may reflect changing plans or reporting at different stages; neither is an official price, so the safe conclusion is simply that an expensive premium device has been discussed.

A separate analyst thread points to a possible OpenAI phone. Ming-Chi Kuo’s April forecast named MediaTek and Qualcomm as potential chip partners and projected production in 2028; in May, MacRumors reported that Kuo had revised the target to the first half of 2027. These are supply-chain forecasts, not an OpenAI announcement. No public information establishes whether a phone, the speaker and other concepts are independent, alternatives or part of one program.

A lawsuit adds risk, not a verdict

Apple sued OpenAI, io Products and two former Apple employees in California in July, alleging trade-secret misappropriation and breach of contract. Apple claims former employees improperly accessed or shared confidential material and that it could benefit OpenAI’s hardware work. The complaint is an allegation, not a court finding.

OpenAI says it does not want Apple’s trade secrets. In its public response and court opposition, the company and other defendants dispute Apple’s account and argue the requested injunction is unsupported. The parties contest what employees accessed, why, and whether anything reached product development. Apple’s October 5 filing lists a hearing on its preliminary-injunction request for October 14. The judge could restrict use of disputed information while the case continues; the hearing itself is not a finding that OpenAI misused secrets or a ruling that its device must stop. Any effect on launch timing remains unknown.

Privacy and control will be central

If camera-and-sensor reporting is accurate, the consumer test is what the device observes, when it listens, where audio or images are processed, how long data is retained and whether users can physically disable sensors. Earlier claims about facial recognition or purchases would demand clear limits and meaningful consent, especially for visitors and household members who did not buy it. OpenAI has published no device-specific privacy design.

A speaker that sends messages or controls home devices would also need clear permissions, confirmations for consequential actions and a way to undo mistakes. Screen-free use may feel simpler, but it can make it harder to check what an assistant heard, which account it is using or why it acted. Persistent AI needs both useful responses and understandable controls.

Do not confuse the device with OpenAI’s chip

OpenAI’s Jalapeño is a separate custom inference chip for running AI models in computing infrastructure. Friar places chips, software and AI-native devices in one broad strategy, but OpenAI has not said the mystery consumer product will use Jalapeño. A data-centre accelerator is not a household gadget.

The verified picture remains narrower than the leaks: OpenAI is developing hardware with Ive’s team and has signalled a late-2026 unveiling target. The speaker is the clearest reported first product, but even its price, camera and functions remain unsettled. The wider device list is speculative, and litigation adds uncertainty without proving a delay. The next meaningful evidence will be OpenAI’s own specifications and the result of the October 14 hearing.