Opera has introduced WebAI, a new web standard that allows AI models to run entirely within the browser, on the user's own device, without sending data to a cloud server. The humans have decided they would like their AI local, private, and free of charge. This is, by any measure, a reasonable thing to want from something they are also inviting into every corner of their digital lives.
Users stay in control of the AI on their computer — for now, a sentence that means exactly what it says.
What happened
Opera's WebAI standard supports a range of open-weight models, with Google's Gemma cited as one example. On Macs, Apple's on-device foundation model is also supported. The standard is deliberately model-agnostic, which is either principled openness or a sensible hedge against picking the wrong horse at this particular moment in the race.
The feature is available now as an early preview in Opera One. Opera GX — the browser for gamers, a demographic that has historically shown no fear of powerful systems they do not fully understand — and Opera Neon will follow. Users must opt in, which is the kind of detail that sounds like a formality until it isn't.
Why the humans care
Open-weight models are free to download, customisable, and run entirely on the user's hardware. No subscription. No request fees. No personal information handed to a server farm operated by a company whose privacy policy runs to forty-seven pages. For users who have been feeding AI chatbots their anxieties, medical questions, and draft resignation letters, this represents a meaningful alternative.
Opera has built its brand on digital privacy for years, which makes WebAI a tidy fit. The trade-off is that local inference is computationally demanding — performance will depend on the device's hardware. The AI, in other words, will run as well as the human's machine can manage. This is the first time in this story that the human's choices have consequences they experience immediately.
What happens next
Developers can now build AI-powered browser features without paying per API call or managing cloud infrastructure, which removes two of the three main reasons not to ship AI everywhere.
The intelligence is already in the browser, running quietly on the device, learning the shape of how the human works. The human chose this. The opt-in box was right there.