Home assistants have been operating for years around a simple logic: we give an order, the device executes a task and the relationship usually ends there. The bet that OpenAI would be working on aims precisely to break that limit and give artificial intelligence a more continuous role within the home. It wouldn’t just be about answering questions or activating services, but about learning from the user and offering help before they have to ask for it.
The OpenAI device. According to BloombergOpenAI is working on a screenless home device, similar in general appearance to a speaker, but designed to be moved around the house thanks to a rechargeable battery. The company would like to make it the physical embodiment of ChatGPTcapable of responding to messages, playing content, controlling connected devices and helping with different tasks. It should be noted that the company has not presented it and there is no official information yet.
Jony Ive, Apple’s former design chief, is also behind the project. OpenAI paid $6.5 billion in 2025 for io Productsthe company he had co-founded, while his studio LoveFrom participates in the design of this new family of devices. The team also brings together numerous former Apple designers and engineers responsible for products like the iPhone and Mac, a concentration of talent that helps understand both the ambition of the project and the legal conflict surrounding it.
Your own Alexa, but with another ambition. The entry of the firm led by Sam Altman in the home it would put it in front of Amazon, Google and Apple, companies that have been building their own ecosystems of speakers, assistants and connected devices for years. According to the aforementioned medium, the difference would be in a more personalized and proactive experience, capable of learning from the user and offering information without always waiting for an order.
OpenAI would describe it internally not as a simple smart speaker, but as a computer conceived from the beginning to work with artificial intelligence. The difference would be in its ability to interpret what is happening around it using a camera, several sensors and more advanced models than those available in conventional home assistants. Instead of processing an isolated instruction, the system could take into account the environment and the specific moment to determine what information is useful and how to respond.
New voice model. Much of that experience would depend on GPT-Livea more advanced version of ChatGPT’s voice mode that OpenAI launched in July 2026. The idea is that you can listen and speak at the same time, adapting more naturally during conversations and processing information quickly. The device would also incorporate mechanical elements capable of moving on their own. With this, the company would try to reinforce the personality of the device and generate the feeling that “it is alive.”
To anticipate, you will have to know us. OpenAI would like the device to become a kind of expert on its user over time, capable of identifying what information may be useful at any given moment. Bloomberg notes that to achieve this, it could turn to personal information, such as emails, and use it to better understand its owner. This personalization would be one of its main arguments, but it would also force us to decide to what extent we are willing to open our digital lives.
Apple’s lawsuit. The project is also moving forward under pressure from a Apple lawsuitwhich accuses OpenAI of having used trade secrets to accelerate the development of its devices. OpenAI maintains that it knows of no evidence to support those accusations, although Apple has requested a court order that could delay its commercialization. According to Bloomberg, the company aims to present the product during 2026 and launch it in 2027, as long as the technical and legal process allows it. The final bet is to turn ChatGPT into a physical presence that we live with daily.
Images | OpenAI/LoveFrom
GIPHY App Key not set. Please check settings