# Gemini AI comes to cars with Personal Intelligence

> Google switches on Personal Intelligence in Gemini for Android Auto from October 20, 2026, and leaves out the EEA, so Spain is not part of the launch.
> By José Maldonado · Artificial Intelligence · 2026-10-21

Google recently announced that Gemini is debuting Personal Intelligence in the car, the feature that cross-references email, calendar and photos to answer without the driver having to explain their own context. The company opens the feature today on Android Auto and will bring it to vehicles with Google built-in in the coming months, according to its official blog.

Google [details in its blog](https://blog.google/intl/es-419/actualizaciones-de-producto/informacion/gemini-presenta-inteligencia-personal/) that: Workspace (with Gmail and Google Calendar), Google Photos, Maps and YouTube, plus the Search history the user already has in their account. With that material the assistant handles compound requests, such as a flight's boarding gate and whether that flight is on time, or returning to the car wash that appears in a photo from last week. None of that works if the user does not first tick the Personal Intelligence box in Gemini's settings.

The important detail is at the foot of the announcement. Google keeps the feature to personal accounts and excludes the European Economic Area and Switzerland, the United Kingdom and Nigeria. Spain is inside the EEA, so the launch does not reach its drivers.

## What changes when the assistant already knows where you are going?

Google sums up Personal Intelligence in two capabilities: retrieving a specific piece of data from an email, a photo or a place saved in Maps, and reasoning across several sources at once. The first saves a search; the second allows chaining questions without repeating the context on every voice turn. On paper it seems minor, but while driving it reduces taps on the screen, which is exactly what the industry has been trying to cut for years.

The example the company repeats is the airport one: the driver asks for the boarding gate and, in the same sentence, whether the flight is still on time. Gemini answers both at once because it has read the confirmation in the email. And all because the assistant used to execute precisely dictated orders; now it can infer the destination from an appointment nobody has named out loud.

The difference from what existed until now is the context, not the voice. Gemini arrived on [Android Auto](https://www.android.com/auto/) in November 2025 and on cars with Google built-in on April 30, 2026, first in English and in the United States.

## How is it switched on and what stays out of the car?

The control remains in the user's hands, at least in theory. Personal Intelligence ships switched off and only activates if someone turns it on in Gemini's settings; if it was already active on the phone, it travels with the account to the car without a second confirmation. After all, Google has a habit of pushing its users toward exchanging data with Gemini, so a box ticked months ago can end up governing the dashboard's answers.

The switch lives one level down in the app's settings, in the connected apps section, where Workspace, Search, Photos and the rest of the services are turned on or off. For anyone who wants to review what they have shared without searching blindly, the Gemini activity setting is what decides how much history is kept, and it is the same small print we analyzed when explaining [why you should not enable activity in the Gemini apps](https://observatorioblockchain.com/en/ai/por-que-no-deberias-habilitar-la-actividad-en-las-aplicaciones-de-gemini/), a setting that becomes more relevant here because the car is shared. Without Personal Intelligence everything else still works, with navigation, music, calls and one-off questions, although the answers stop being tailored to whoever is driving.

Cars with Google built-in, with Google's software inside the infotainment system rather than projected from the phone, have an extra condition: they only give personalized answers when the phone is connected in the vehicle.

## Spain is left out of the launch

The rollout has a footnote that changes the scope of the announcement. Personal Intelligence is available on personal accounts of users over 18 and its reach is global with four explicit exclusions: the European Economic Area and Switzerland, the United Kingdom and Nigeria.

The exclusion is not new nor exclusive to the car. Gemini's help page already describes the feature as a beta for eligible accounts worldwide except those same geographies, and places the initial rollout in the United States last March, with the rest of the markets still in beta. It is also not available on enterprise or education Workspace accounts, so the launch stays in the domestic sphere. For a driver in Madrid, the car keeps navigation, music and calls, but the answers arrive without personal memory and without that shortcut that saves looking up the address by hand.

Europe watches these launches with the artificial intelligence regulation in hand, a calendar of obligations that conditions how the continent's startups design their features, as we reviewed when analyzing the [AI Act and European startups](https://observatorioblockchain.com/en/ai/ai-act-startups-europeas-inteligencia-artificial/).

## Android Automotive versus CarPlay, two ways of getting into the dashboard

Android Auto and CarPlay share a logic: the phone runs the software and the car screen merely projects it. Google built-in breaks that scheme because the system lives inside the vehicle, starts without a phone and gives access to native apps. Google recalls that those cars reached the market in 2020 and that its commitment was to improve them with updates for years, not just to sell them with the software of the day of delivery.

Via

Where the software runs

Personal Intelligence

Android Auto

On the phone, projected to the car

Since October 20, 2026

Google built-in

In the vehicle's system

In the coming months

CarPlay

On the iPhone, projected to the car

Not applicable, the assistant is Siri

The paradox of the announcement is that Google usually debuts its car features first on Google built-in, and this time it has done the opposite: Android Auto, the path that depends on the phone, gets Personal Intelligence before vehicles with the integrated system. Brands such as Ford, GM, Honda, Nissan, Volvo and Polestar sell cars with Google built-in, although very few vehicles carry the integrated system. The car screen stops being an accessory and becomes another surface where the user confirms, without thinking about it, that their calendar and their email live in the same cloud.

There lies the fundamental difference between the two strategies. Apple keeps the phone as the brain and charges for the hardware; Google can give the assistant away on the dashboard because the value is in the accounts it already manages.

## The price of giving context away to the dashboard

Google does not hide the commercial argument: the more a model knows about the user, the better it responds. This can be summed up in one sentence: it is an assistant that works better when it knows the world and knows you. The uncomfortable nuance is that this context was already inside the account, so the advantage over other assistants does not come from a superior model, but from a personal archive the user built without thinking about the car.

The same outlet adds two clarifications worth retaining. Google maintains that it does not train Gemini with personal data, but it does use the assistant's inputs and outputs, and some of those texts may contain private information when services are connected; the company tries to filter it and does not promise to always succeed. The result is a better-informed assistant and a trail that is harder to audit, with the added fact that the car usually has passengers.

The debate about an artificial intelligence that does not depend on a single provider's cloud already has a name of its own in the ecosystem, and the proposal we described in [Neither ChatGPT nor Gemini: this is the self-sovereign AI proposed by Vitalik](https://observatorioblockchain.com/en/ai-agents/ni-chatgpt-ni-gemini-asi-es-la-ia-autosoberana-que-propone-vitalik/) points in exactly the opposite direction to this announcement. There the context stays on the device and the user decides what to share; here the context is the product.

While the European Union remains outside the rollout, the rest of the market grows accustomed to an assistant that knows the route before the driver asks for it. When that feature crosses the Atlantic, the discussion about what data enters the dashboard will already be settled in favor of whoever holds it.
