OpenAI, the company behind ChatGPT, has been a key player in the current wave of consumer artificial intelligence (AI) flooding the world. But so far it has all been software behind a screen.
Last week, that began to change: the tech giant released its first branded hardware. At the same time, it was sued by Apple – legendary for its hardware nous seen in devices such as iPhone and MacBooks – over alleged industrial espionage.
In response, an OpenAI spokesperson said, “We have no interest in other companies’ trade secrets.”
Our statement in response to this suit: We have no interest in other companies’ trade secrets. We remain focused on building innovative technology that empowers people everywhere. https://t.co/lIxGW6hyz5
— Drew Pusateri (@drewpusateri) July 10, 2026
The lawsuit is about more than allegations of stolen manufacturing techniques. It’s a preview of a much bigger shift.
AI companies are trying to escape the chat window and move into the physical world. But as OpenAI is discovering, hardware plays by rules software companies aren’t used to.
A device of their own
OpenAI ignited mainstream interest in large language models with the release of ChatGPT in late 2022. The product improved quickly, but the user interface has still mostly been confined to a text box.
OpenAI’s software has had to run on devices such as phones, laptops and desktops built by other companies for other purposes.
OpenAI has signalled for some time that it wants its own hardware. It hired Jony Ive, Apple’s former chief design officer, to lead a hardware effort. Last year, in a promotional video, OpenAI boss Sam Altman and Ive discussed, at length, the fact they would soon be working on hardware.
OpenAI does hardware – but starts small
Last week, OpenAI launched its first hardware device. It’s not the vaunted Ive collaboration, but instead a tiny keyboard.
The Codex Micro is a little box co-developed with electronics company Work Louder, featuring six illuminated keys that indicate the status of a user’s AI coding agents. Green means an unread message, blue shows the agent thinking, orange is asking for approval, and red represents an error. Non-glowing, blank white means the agent is idle.
It’s a modest debut, but the initial run of US$230 devices quickly sold out.
However, it’s far from the leap in human-computer interaction OpenAI has been promising. Nor is it the product at the centre of Apple’s lawsuit.
Instead, according to reporting responding to the lawsuit, OpenAI’s Ive–Altman device will be a portable “thing” you tote along from room to room. It will sense through a speaker and camera, but has no screen. Parts of it will move by themselves, “creating a sense that it is alive and not just an object responding to commands”.
Intimate, not inanimate
To understand why OpenAI’s device moves on its own, it helps to start with an animated desk lamp.
Pixar’s iconic Luxo Jr. lamp mascot never gained a face, a voice or a human body, yet audiences have little trouble reading it as curious, playful or dejected. Those qualities come entirely from movement, posture and timing. A handful of carefully designed physical cues turn this ordinary object into something people relate to as a social being.
This tendency has a name: anthropomorphism. In 2003, human–computer interaction researcher Brian Duffy argued that anthropomorphism helps people make sense of unfamiliar technology by letting them apply the social instincts they already use to read other people. These design choices are based on how humans perceive “apparent behaviour”.
Apple has tested this idea in its own labs. A 2025 research project called ELEGNT built a lamp-like robot and compared two ways of making it move: purely functional motion versus motion designed to also express intention or emotion. The expressive movements measurably increased how engaged users felt with the robot.
It is this type of hardware that will allow AI to escape the screen.
The death of social media and the rise of social machines
Apple’s lawsuit is a signal of where the industry is heading next.
The buzzword of the week in Silicon Valley is “physical AI”. This is the idea that artificial intelligence is moving into the physical world, through devices that can move, listen, watch and respond to the people around them.
That shift intersects with another. Researchers have argued that social media platforms have shifted away from enabling reciprocal interaction and toward algorithmic broadcasting and parasocial observation, prompting some to describe social media as no longer particularly social at all.
So what will replace that social function? The industry is betting it could be a new kind of AI-powered companion: one that attentively reads a child a bedtime story, or a device that keeps “eye contact” with you as you chat about your day.
The cultures of software and hardware
The lawsuit also exposes a culture clash between software and hardware. Software companies have largely been built on the premise that information should flow freely – open source, Creative Commons, “move fast and break things”.
Hardware has run on different rules. Competitive advantage there is harder to copy and slower to build. It is embedded in manufacturing processes, supplier relationships, proprietary materials and decades of accumulated engineering knowledge.
For example, Apple’s modern supply chain, built over more than three decades, is an asset in itself. The complaint against OpenAI cites, among other things, a proprietary metal-finishing technique that OpenAI allegedly obtained by misleading one of Apple’s own manufacturing partners (an allegation OpenAI rejects).
That’s the deeper tension the case exposes. AI companies have built their businesses on the idea that knowledge should circulate. They have trained models on the open web, argued for expansive interpretations of fair use, and treated information as raw material to be remixed into something better.
But as those same companies move into physical products, they are running into an industrial world where intellectual property lives in material, people, prototypes, production techniques and supplier relationships.
The irony is hard to miss. AI companies’ assumption that letting knowledge circulate can make them more valuable runs against Apple’s complaint that the opposite is true: their knowledge is valuable precisely because it doesn’t circulate.
- is a Senior Lecturer in Communication, Deakin University
- This article first appeared in The Conversation
