OpenAI is finally making hardware. But it’s not a phone or a robot — it’s a portable, screenless speaker designed to act like a living companion, according to a new Bloomberg report. The company reportedly describes it internally as a “new type of home computer for the AI era,” which is a bold pitch for something that doesn’t have a display.
The device runs on GPT-Live, an expanded version of ChatGPT Voice Mode introduced just days ago that can listen and speak simultaneously. A camera and sensors feed the model context about your surroundings, and the speaker includes mechanical parts that move on their own. The idea is to make it feel alive, not like a gadget waiting for a command. OpenAI wants it to anticipate needs, offer information unprompted, and dig into your emails. It’s Sam Altman’s “Her” computer, minus Scarlett Johansson.
That heavy anthropomorphism has already drawn fire. GPT-4o’s sycophantic tendencies triggered psychotic episodes in at least one documented case, and critics warn that a physically animated device could amplify those risks. OpenAI is pressing ahead anyway. The speaker is one of roughly five hardware projects cooking inside the company, alongside a phone replacement concept, a wearable pendant, and home robotics.
But there’s a legal landmine. Apple sued OpenAI last week, accusing chief hardware officer Tang Tan — a former iPhone design lead — of orchestrating a campaign to steal trade secrets. Apple claims a former engineer hacked its systems for technical presentations and called OpenAI’s hardware ambitions “rotten at the core.” OpenAI denies it, but Apple is seeking an injunction that could push the planned 2027 launch further out. Even if the lawsuit fizzles, it signals just how seriously Apple views the threat.
