OpenAI president Greg Brockman confirmed the company is building a family of devices designed for interacting with its AI models, and he thinks you’ll talk to them more than you’ll type. He confirmed these OpenAI devices are coming and form a product family. In an interview with Joanna Stern published Tuesday, Brockman laid out a hardware vision that goes beyond any single gadget. It’s a bet on voice-first computing as the default way humans interact with AI.
Brockman didn’t confirm the rumored smart speaker or the Jony Ive wearable that’s been speculated about for years.
But he did say OpenAI’s devices are coming soon, that they form a family of products, and that the company is building its own chips to power them.
Why voice matters more for OpenAI devices
The biggest signal in the interview is how Brockman thinks computing will work. He said he believes we’ll shift to talking to our computers rather than typing for the vast majority of what we do. That’s a massive claim from the president of the company building the most popular AI chatbot on the planet.
When Stern asked about the obvious problem (everyone in an open office talking to their computers at the same time), Brockman pointed to soundproof beak masks that have started appearing. He said he doesn’t use one himself, but called them proof of product-market fit for a voice-first interface, even if everyone agrees it’s not a great long-term solution.
It’s hard to argue with the direction. I’ve been saying for months that the keyboard-and-mouse paradigm is the main thing holding back how useful AI agents can actually be. Typing a prompt, waiting for a response, reading it, typing the next correction. That’s not interaction. That’s slow-motion Q&A. Voice removes that friction, and OpenAI knows it.
Apple lawsuit isn’t slowing them down
Brockman was direct when asked about Apple’s trade secret lawsuit against OpenAI, which accuses the company of stealing hardware design information from Apple employees who were interviewing with OpenAI. “We have no interest in other companies’ trade secrets,” he said. “We are plenty innovative. We are thinking about things from our own angle.”
The lawsuit specifically targets the hardware OpenAI is building with former Apple designer Jony Ive, though Ive isn’t named in the complaint. Apple described OpenAI’s hardware efforts as relying on stolen information, but Brockman’s response is essentially: we don’t need your ideas.
ChatGPT itself, when asked what OpenAI’s first device would look like, described it as “ambient and low friction, voice first but not voice only, and with a clear privacy signal.” When asked whether the device would need a screen, ChatGPT said “nope.”
The ChatGPT app is “kind of a mess”
Brockman also addressed the state of the ChatGPT desktop app, which recently merged with Codex into ChatGPT Work. He acknowledged it’s “kind of a mess” and described the current app as incremental progress toward the future. The separate Work tab should be gone by the end of the year, with everything folding into ChatGPT.
That’s good to hear. The app has felt like two products taped together since the Work merge, and power users have been vocal about it. A unified experience is what the product should have been from day one.
Privacy guarantees are coming
Brockman also said OpenAI is working on trust issues with its models, including concerns about how private ChatGPT’s temporary chat mode actually is. He mentioned developing technological solutions that offer “verifiable, auditable guarantees” that only the AI can review users’ private data.
That’s the kind of commitment that matters for anyone who’s hesitated to put sensitive information into a chat interface. Hardware without trust guarantees would be a non-starter, so this tracks with the device push.
What this means
OpenAI moving into hardware is the kind of vertical integration play we’ve seen from Apple, Google, and Amazon. Control the chip. Control the device. Control the AI.
It’s a strategy that says the current era of using ChatGPT through a browser or an app is a stepping stone, not the final form.
The smart speaker rumor makes the most sense as a first product. An ambient, always-listening, screenless device that runs OpenAI’s models directly fits every clue Brockman dropped. It’s also the hardest product to get right. Voice interfaces that work well in noisy, real-world environments are still a genuinely hard problem.
“Expect them soon” is vague, but it’s more specific than anything OpenAI has said publicly about hardware before. Combined with the chip work and the family-of-devices framing, this is the clearest signal yet that OpenAI plans to own the full stack from silicon to interface.
Bottom line
OpenAI confirmed it’s building hardware, and that hardware is voice-first, screen-minimal, and coming sooner than most people expected. The bet is that talking to AI will replace typing for most tasks, and the company is building the chips, the devices, and the privacy infrastructure to make that work. The ChatGPT desktop app is a mess right now, but that’s because it’s transitional. The real product hasn’t arrived yet.
