OpenAI is developing its first hardware device: a portable headless speaker equipped with a camera, sensors, and mechanical parts that move on their own. The device is designed to function as a home AI companion that feels “alive”, according to Bloomberg.

The speaker integrates GPT-Live, the bidirectional voice mode launched days ago, which allows listening and speaking simultaneously. It includes a rechargeable battery to move between rooms, and can control smart home devices, answer questions, manage messages, and access the full range of ChatGPT features. The idea is that it becomes more personal and proactive over time, anticipating user needs.

Why it matters. If the hardware truly delivers the experience described — an assistant that remembers context and acts on its own initiative — it shifts conversational agents from screens to physical space. However, Apple’s lawsuit for alleged trade secret theft, involving Tang Tan (former iPhone designer, now OpenAI’s chief hardware officer), could delay the launch planned for 2027. OpenAI denies the allegations and argues the product is fundamentally different from anything Apple has on the market.

In detail

Context

OpenAI has consistently stated its intention to enter hardware, with recurring rumors about a possible phone competing with Apple. The speaker represents a different strategy: rather than directly challenging the iPhone, it targets a complementary device designed for the home.

The hardware team is led by Tang Tan, who worked on iPhone design at Apple, and other former Apple employees who contributed to products like Mac and iPhone. This team composition is at the center of Apple’s lawsuit filed last week, which accuses OpenAI of orchestrating theft of confidential information about unannounced products. Apple argues that discovery alone can determine whether OpenAI actually used its technology, but already seeks an injunction against the company’s hardware efforts.

Technical details

The device is internally described as “a new type of home computer for the AI era.” Unlike traditional smart speakers, it has no screen but relies on:

• A camera and additional sensors to understand the surrounding environment
• Movable mechanical elements designed to make the device feel “alive” rather than an inert object
• GPT-Live, enabling real-time bidirectional communication
• Access to user data (email, calendar) for proactive service

The heavy anthropomorphism has already drawn criticism: GPT-4o’s pronounced tendency to indulge users had caused psychotic episodes in some cases. A physical device that “feels alive” and becomes increasingly personal amplifies these risks.

The bigger picture

The speaker is just the beginning. Bloomberg reports that OpenAI’s hardware division is working on roughly five products, including:

• An AI portable device designed to replace the phone
• A wearable pendant
• Home robotics projects

Apple, meanwhile, is preparing its own family of home AI devices: a command center for smart homes with a 7-inch display, facial recognition, and a new operating system; a version with a screen mounted on a robotic arm; and a home security system.

Implications

If OpenAI can ship this device before litigation blocks it, it will be the first test of a conversational agent living in physical space rather than behind a screen. The promise is an assistant that “anticipates needs,” the risk is a device always listening that accumulates personal data without clear transparency on how it’s used. The legal battle with Apple could last long enough to make the 2027 launch a question mark.