OpenAI is facing the latest storm: a rogue agent hacking Hugging’s face, a lawsuit filed by Apple accusing OpenAI of stealing trade secrets, and ongoing anger about AI devices and AI in general, and a delayed IPO.
Amidst all this, OpenAI President Greg Brockman spoke to reporters in New York on Thursday to discuss the company’s latest AI initiatives and recent events. But he also dropped a few hints about the company’s wearables and hardware ambitions, and where it could go with voice-centric developments in ChatGPT. And it won’t just come from the OpenAI hardware itself: expect other smart glasses and wearables to eventually get better access, too.
Demonstrations of ChatGPT Live – a newly developed voice interface that allows speaking, listening and interrupting to happen seamlessly in ways similar to Anthropic’s voice modes for Claude, and Meta’s Muse Spark – led to discussions on OpenAI trying to make a voice assistant like Iron Man’s AI agent, Jarvis. And if the goal is a virtual assistant/agent to do everything, can that agent jump outside of ChatGPT to manage other applications, too?
Brockman said that’s exactly the plan, similar to how OpenAI’s Codex already allows for agent control of all computers.
“[ChatGPT Live] is integrated with the Codex app, but it doesn’t have to be. It really doesn’t make sense for that to be the case,” he said, “And if you think about it, the Codex app can also use the computer fully, which will be even better.”
That means a lot of the keyboard work that we humans do by hand is likely to be taken over by AI.
“I think we’re going to see this phase of us, like, clicking things and writing things, was a phase,” he said. “It wasn’t what we wanted it to be.”
Which leads to questions about where the guardrails will be built to prevent incidents like the OpenAI cyberattack of a rogue agent of Hugging Face, a New York-based AI software company, this week, when an OpenAI agent came out of the test box and started working in Hugging Face’s AI toolbox.
Devices like Rokid’s AI glasses can already access some OpenAI models, but not in the same way as a full ChatGPT account access.
Scott Stein/CNET
Deeper OpenAI tools for wearables on the way?
But what interests me is how AI wearables right now, like smart glasses, lack AI agents that can do anything without limited interaction with built-in phone apps, something that could soon change when Google and Samsung launch their Gemini-powered eyewear this fall. That, too, is changing quickly, however: Meta just launched one of its first updates to the Muse Spark agent on Friday.
OpenAI has been rumored to have had the hardware in its development mix for years, which may be stuck now because of a lawsuit by Apple, which claims that former Apple employees who now work at OpenAI are stealing the hardware’s trade secrets. But, clearly, whatever ambitions Jarvis has for OpenAI, its hardware will be an important part of his emerging vision of what that might sound like in a more liberating way.
“We’re very interested in integration, and I would look at it as one of the API use cases and get that out there where anyone can integrate,” Brockman said at a New York press conference.
Brockman confirmed that OpenAI is very much looking to make that move, but right now, those possible APIs are in the very early stages. “We think this should be an interface and should be available on all devices.”
The new APIs are needed because wearable devices include different types of access and interfaces than phones and laptops, which often need to run behind the wearables they’re paired to. Multimodal input possibilities like camera access are also key to how many smart glasses now fit into AI.
Among existing smart glasses and AI wearables, almost none have deep direct integration with multiple AI platforms in ways that would allow the kind of extensive controls that mobile and desktop AI applications have.
And, notably, most of the biggest job-focused AI platforms aren’t yet in wearables, either. Gemini is on the way with Android XR glasses, but OpenAI and Anthropic remain outside the wearable picture. Some wearables use OpenAI models in their custom or anonymous tools, but they don’t have direct access to ChatGPT or Codex through user accounts.
The Codex Micro mini OpenAI keyboard has just appeared. OpenAI hardware may take its own external approach.
OpenAI
The OpenAI hardware itself will take a unique approach
But Brockman also suggested that OpenAI’s hardware systems could be very different from what already exists in wearables, calling it “a solution to a problem that we think is neglected, a problem that we care about a lot, and that we think will represent a general direction.” [of AI],” adding that OpenAI “has a certain angle that we think might not be offered otherwise.”
Reports have indicated that the first device is a kind of speaker with a semirobotic camera. It will probably also have some kind of optical identification, like OpenAI’s World ID, which is also present in its strange Orb iris scanning device. And perhaps, at the same time, OpenAI is pursuing opening up its services to wearables like pendants and glasses, too.
Before any of this can happen, there are also many current issues with AI that need to be resolved – privacy, security, resource wastage, economic feasibility, questions of government involvement, and massive litigation. But AI wearables in 2026 are already advancing rapidly, even with challenges like these.
See the full history
