General Motors is bringing Google’s Gemini AI to approximately four million Chevrolet, GMC, Buick, and Cadillac vehicles through an over-the-air software update rolling out in spring 2026. The upgrade, which layers Gemini’s large language model on top of the Google built-in system already installed in models from the 2022 model year forward, is one of the largest single deployments of generative AI into cars already on the road.
The rollout lands just months after the Federal Trade Commission finalized a settlement with GM and its OnStar subsidiary over allegations that the company collected and sold precise driver geolocation data without informed consent. That sequence puts a spotlight on a question millions of GM owners are now facing: what happens to the data generated inside their cabins once a more powerful AI is listening?
What the Gemini upgrade actually changes
Drivers who already use the Google Assistant voice interface in their vehicles will notice the shift the next time they start their car after the update downloads. According to GM’s announcement, Gemini replaces the underlying language model powering the assistant, enabling more conversational, context-aware responses. Instead of rigid command-and-response exchanges, the system is designed to handle open-ended questions: asking for a dinner spot that fits a dietary restriction near an upcoming destination, for example, or requesting a summary of traffic conditions along a planned route.
The update also expands the assistant’s ability to control vehicle settings through natural language. GM says drivers can adjust climate, lighting, and media playback with less precise phrasing than the current system requires. Eligible models span a wide range, from the Chevrolet Silverado and Equinox EV to the GMC Sierra, Buick Enclave, and Cadillac Escalade, among others.
What GM has not detailed publicly is the technical boundary between on-device processing and cloud-based queries. Basic commands like adjusting fan speed may stay within the vehicle’s hardware, but requests involving nearby businesses, real-time traffic, or points of interest almost certainly require data to travel to external servers. Google’s automotive privacy documentation describes data collection in broad categories without specifying what Gemini accesses during a given session or how long query data is retained.
The FTC settlement and why it matters now
The regulatory backdrop is impossible to ignore. In January 2026, the FTC finalized its order against GM and OnStar in Matter No. 2423052. The formal Complaint alleged that GM gathered precise geolocation data from connected vehicles and sold it to third-party data brokers and other companies, all without obtaining meaningful consent from drivers.
The resulting Decision and Order imposes binding restrictions. According to the FTC’s January 2026 press release, the agency’s then-Chair Lina Khan stated that the order “bans GM and OnStar from sharing consumers’ precise geolocation data and driving behavior information with consumer reporting agencies” and requires the companies to “obtain consumers’ affirmative express consent” before collecting or disclosing such data. The order also includes broader limits on data practices and new consumer choice requirements. Violations carry financial penalties.
These restrictions apply to OnStar, the connected-vehicle platform that has served as GM’s data backbone for more than two decades and the same infrastructure now being layered with Gemini’s capabilities. That overlap raises a practical question the settlement documents do not directly address: does an AI assistant that uses real-time location to generate personalized suggestions constitute the kind of data sharing the order was designed to prevent, or does it fall into a different category of “processing” that the FTC’s language does not explicitly cover?
Neither GM nor the FTC has publicly clarified that boundary. The investigation predates the Gemini integration, and the Complaint and Decision and Order, as published on the FTC’s case page, do not reference AI-specific data processing scenarios.
What GM has not said
Several gaps in GM’s public communications stand out. The company has not disclosed whether the FTC settlement prompted any changes to the Gemini rollout timeline or its technical design. No public filings or on-the-record statements from GM connect the two developments. The proximity in timing, with the AI deployment following months after the finalized order, could reflect behind-the-scenes adjustments or could be entirely coincidental. Without internal documents or statements from GM’s legal or engineering leadership, drawing a causal link remains speculative.
GM has also not published a technical audit or detailed privacy impact assessment specific to the Gemini integration. Its announcements describe features and user benefits but stop short of explaining data flows, retention timelines, or how consent dialogs will be structured for AI-powered queries that rely on location. Google’s automotive partnership materials follow a similar pattern: heavy on capability descriptions, light on data governance specifics.
Consumer privacy organizations have raised general concerns about AI in connected vehicles. The Mozilla Foundation, which has previously rated automakers poorly on privacy in its “Privacy Not Included” reports, noted in its 2024 automotive review that “cars are the worst category of products for privacy that we have ever reviewed.” But as of May 2026, no published independent research specifically evaluates Gemini’s privacy architecture inside GM’s fleet. That leaves drivers relying on corporate disclosures and regulatory enforcement rather than third-party technical assessments.
How competing automakers are approaching in-car AI
GM is not operating in a vacuum. Ford has expanded its partnership with Amazon’s Alexa for voice controls, and Mercedes-Benz has integrated a ChatGPT-powered assistant into select models through its MBUX system. Stellantis has signaled plans for AI-driven features through its STLA platform. But none of these competitors has attempted a Gemini-scale generative AI deployment across a fleet as large as four million vehicles simultaneously, which makes GM’s rollout a de facto test case for the industry.
It also makes GM the first major automaker to push generative AI into millions of cars while operating under an active FTC data-privacy order. How the company navigates that dual reality, delivering a competitive AI product while demonstrating compliance with binding consent requirements, will likely influence how regulators approach similar integrations from other manufacturers.
What GM owners can do right now
For drivers who want to manage their exposure, several concrete steps are available. Reviewing privacy settings in both the vehicle’s infotainment system and any linked Google account is the most immediate action. Turning off location sharing where the option exists, declining optional data-sharing prompts, and limiting personalized recommendation features can reduce the volume of information transmitted beyond the vehicle.
Owners should also watch for new consent dialogs that appear after software updates, particularly those referencing navigation, recommendations, or connected services. These are likely the touchpoints where the FTC’s opt-in requirements are being implemented. Agreeing reflexively, as most people do with software prompts, could authorize data flows that the settlement was specifically designed to put under driver control.
It is also worth distinguishing between features that require cloud connectivity and those that do not. Voice commands for climate or media playback may function largely on-device, while queries about nearby restaurants, traffic, or route planning will almost certainly involve server communication. Where connectivity is essential, some data transmission is unavoidable, but drivers can still choose whether to tie those services to a personal Google account or use them without signing in, where that option is available.
A regulatory template still being written
The FTC’s order against GM offers one of the earliest regulatory frameworks for connected-car data practices, but it was crafted before generative AI dashboards became a commercial reality. Whether its consent requirements and data-sharing bans prove flexible enough to govern an AI assistant that processes location, voice, and behavioral data in real time is an open question that regulators, automakers, and privacy advocates will be working through for years.
For the four million GM owners now receiving Gemini, the practical trade-off is immediate: a more capable, conversational in-car assistant in exchange for deeper integration with cloud-based AI systems whose data practices are not yet fully transparent. The FTC settlement ensures that certain lines, especially around selling precise location data, cannot be crossed without explicit permission. But it does not resolve every ambiguity about how a large language model will handle the information generated inside a moving vehicle.
Drivers who value the convenience may find the upgrade worth embracing. Those who remember why the FTC investigated GM in the first place may want to read the consent prompts carefully before they tap “Agree.”
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
