I use a Galaxy S23 as my primary phone, and there is one Galaxy AI setting I always turn on. It is called "Process data only on device", and it does what it says. When the setting is turned on, my Galaxy S23 runs only the AI features that work locally, and nothing is sent to Samsung's servers. After researching this week at how Google splits Pixel AI features between the on-device AI and cloud AI, I kept thinking Google should copy it.
What This AI Toggle Actually Does
The toggle is available under the Galaxy AI menu in Settings. You can turn on "Process data only on device", and Galaxy AI is limited to the features that can run on the phone only. For instance, AI features like live call translation, call screening, chat translation, writing tools in the Samsung Keyboard, voice transcription and photo eraser are all accessible with this setting turned on.
Sure, you need to download a language pack first for some features, but once it's there, the work happens entirely on the phone. On the plus side, the heavier AI features stop working, as they need Samsung's cloud AI models. That includes generative photo edits, sketch to image, Portrait Studio, web page summaries in the Samsung browser and more.
What I like is that when you tap a cloud-based AI feature while the toggle is turned on, the phone does not just stay silent. It tells you the feature needs cloud processing and asks you to turn off the toggle first. This way, you never send data to a server by accident because you actively opt in to cloud processing.
Related Articles
This Is Exactly What Pixel Is Missing
Google talks a lot about on-device AI, especially with the Tensor G6 chip and its new TPU, but it does not clearly tell you which of the Pixel 11 AI features run on the phone and which ones go to the cloud.
I spent a good chunk of this week trying to categorise on-device AI features and cloud AI features on the Pixel 11. Even then, some features use a vague hybrid model, and Google does not clearly spell it out. A normal user is never going to do that research.
On the other hand, Samsung just hands you a one-stop toggle. In fact, you can tap the setting instead of the toggle, and you get a list of all Galaxy AI features which work offline using on-device AI models. From Daily brief to Notification highlights to Audio eraser, you get most of the Galaxy AI features locally.
As you can see, it's a long list, which shows how much AI can already run on a phone without touching the cloud. That alone is more transparent than what Google does, and I am happy that at least Samsung offers this option.
Related Articles
Google Should Be Confident Enough to Offer This
Of all the companies, I think Google should be the one doing this. It designs its own Tensor chip, it builds the TPU that runs Google's own Gemini Nano model, and it keeps telling us that private, personal AI is the future. A company taking that position should be happy to allow users to use AI services transparently.
A toggle like this would also fix the problem of consent around AI use. Right now, AI features are turned on by default, and you often can't tell where your data goes. With Samsung's approach, you effectively opt out of cloud processing, and you only opt in when you tap a feature and agree to cloud processing. The user stays in control, and the phone informs you about the trade-off.
That said, Samsung's toggle only covers its own Galaxy AI features, so the Gemini app still uses the cloud. It's not perfect, but it is far better than having no option at all, and it's the kind of thing Google could build properly across the whole Pixel experience.
So this is my simple request for Pixel. Give me one clear toggle that keeps AI on-device, show me which AI features are covered under on-device AI and prompt me before any data leaves my phone.
Related Articles
-1120w630h.webp "Vivo X500 Pro Confirmed to Get 2nm Dimensity Chipset with Major Video Upgrades thumbnail")
