General Motors plans to roll out an in-vehicle AI assistant later in 2026 that taps into OnStar data to understand your car’s health, settings, and service needs—going far beyond the conversational skills of Google Gemini, which is already arriving in millions of vehicles this year. First reported by CNBC, the unnamed assistant represents a strategic bet that car owners will pay for an AI that can do things like predict maintenance, coordinate child-friendly settings, and interpret telemetry in plain language.
GM’s AI Roadmap Splits Into Two Lanes
GM has been unusually transparent about its plan to layer two AI systems into its vehicles. You’re already getting the first one: Google Gemini.
Starting this summer, Gemini will land in approximately 4 million eligible vehicles in the U.S.—any 2022 or newer Cadillac, Chevrolet, Buick, or GMC model that has Google built-in. The rollout, delivered over several months via over-the-air update, upgrades the existing Google Assistant to a much more natural, context-aware chatbot. You can ask it questions about landmarks, draft texts with emojis, summarize podcasts, or create a road trip playlist that matches the mood and the mileage. It can handle follow-up questions, translate messages on the fly, and even juggle multiple tasks in a single sentence.
But Gemini’s job is to be a general-purpose assistant. It lives largely in the infotainment layer—it knows your media apps, your contacts, and the map, but it doesn’t really know your car. That’s where GM’s own assistant comes in.
According to CNBC, GM is working with an undisclosed large-language-model provider to build a second assistant that will be “fine-tuned with vehicle intelligence and personal preferences through OnStar.” GM’s own press materials describe it as a system that can anticipate your needs by connecting to the vehicle’s telemetry. This isn’t a chatbot that just answers questions; it’s a co-pilot that can read diagnostic codes, track your driving patterns, and suggest service before a part fails. A single command like “kids setting” could auto-configure the climate, lock the rear windows, pull up a kid-friendly playlist, and even remind you to check the back seat—all because the assistant understands the vehicle’s configuration and your past behavior.
The timeline is still fuzzy. GM has only said the integrated assistant will arrive “later in 2026,” with no specific month, no named vehicle models yet, and no word on whether it will require a paid subscription. But the company has made one thing clear: Gemini remains the front door for conversational AI, while its own system is the backstage engineer.
Why OnStar Data Changes the Game
The secret sauce isn’t just an LLM; it’s the 30 years of connected-car data flowing through OnStar. GM has over 16 million connected vehicles collecting everything from oil life to exact coordinates. That data lake lets the company train an assistant to be vehicle-specific in a way that a third-party AI cannot.
Think about the difference between asking Gemini “Is my car due for service?” and asking GM’s assistant the same thing. Gemini might give you generic advice about oil change intervals. GM’s assistant will pull the actual odometer reading, cross-reference it with your driving habits, and tell you that the brake pads are at 4mm and should be replaced within the next 1,000 miles—and then offer to schedule an appointment at your preferred dealer.
This deep integration also opens the door to more complex automation. A fleet manager could query the assistant to check which of 50 trucks have low tire pressure, or to route a vehicle to the nearest charging station based on real-time battery range and cargo weight. For everyday owners, it could mean never having to decipher a dashboard warning light again.
But the dependence on OnStar raises immediate privacy questions. The assistant will need continuous access to vehicle telemetry and possibly location data. GM hasn’t yet explained how it will handle data retention, whether processing will happen locally or in the cloud, or how owners can opt out without losing functionality. And once an AI can control multiple cabin systems—locks, windows, climate—security becomes paramount. An over-the-air misconfiguration or a compromised LLM prompt could have consequences far beyond a wrong playlist.
What This Means for You (and Your Privacy)
If You Drive a Recent GM Vehicle
Good news: Gemini is free for now, assuming you keep your OnStar connection active and log into the Google Play Store. Over the coming weeks, you’ll see a notification on your infotainment screen when the update is ready. After opting in, you can start using natural language commands immediately. Try asking for a summary of the latest news, or tell Gemini to “find a coffee shop along my route and text my spouse that I’m stopping.” It’s a genuine step up from the old Google Assistant.
The bigger payoff, though, is still months away. If your car is among the first to receive the GM native assistant later in 2026, you’ll get an AI that knows your vehicle’s service history and can tailor the driving experience. Whether you’ll have to pay extra for that remains unknown. GM has a history of bundling OnStar services, so it’s plausible the assistant could be part of a premium connectivity plan.
If You Manage a Fleet
The dual-AI strategy has practical implications for fleet operations. Gemini can already help with route planning and fuel cost optimization, as GM’s commercial-vehicle demos show. But the real value for fleet managers will come when the integrated assistant can monitor vehicle health across an entire fleet, predict breakdowns, and automate maintenance scheduling. However, you’ll need to wrestle with data governance: who at your organization has access to the assistant’s insights? Are conversations logged? How do you ensure that sensitive business data isn’t fed into an LLM that might retain it? GM will need to provide enterprise-grade controls before IT departments are comfortable deploying this.
If You Care About Privacy
The assistant’s promise is built on access to your vehicle’s data. GM has been collecting OnStar data for years, and its privacy policies allow sharing anonymized information with third parties. An AI that personalizes based on your driving patterns will inherently need to store some of that data. Before you opt in, check your OnStar privacy settings at gm.com/privacy. Right now, you can mitigate data sharing by disabling certain data collection options, but that might also disable future AI features. When the integrated assistant launches, read the fine print carefully: does it process commands locally on the vehicle’s hardware, or are your words and vehicle status sent to the cloud? The difference matters.
The Road to Smarter Cars
GM didn’t arrive at this point overnight. The foundation was laid in 1996 with the first OnStar system, which could automatically call for help after a crash. Over the decades, OnStar evolved from a safety net into a full connectivity platform, enabling remote start, vehicle diagnostics, and Wi-Fi hotspots. That made it possible for GM to push software updates fleetwide, a capability that rivals only Tesla’s at scale.
The company’s AI ambition became public at the GM Forward event in 2025, where executives teased a “custom-built AI assistant” trained on proprietary vehicle data. In April 2026, GM announced the Gemini rollout, with Tim Twerdahl, VP of product management, explicitly stating that a “more deeply integrated AI experience shaped by OnStar intelligence” would follow later that year.
The CNBC report in late July 2026 filled in key details: the assistant is real, it’s coming this year, and it’s being built with an outside LLM provider—though who that is remains hidden. Google is the obvious candidate, but GM’s insistence on using OnStar data for differentiation suggests it might be keeping its options open, maybe even training a custom model on its own infrastructure. Hyundai and Mercedes have also been exploring in-house AI, often in partnership with multiple tech firms. GM’s approach—layer a premium, brand-specific AI on top of a mainstream one—could become the template for other automakers who don’t want to cede all the data to Silicon Valley.
Actions to Take Before the 2026 Launch
- Check your vehicle’s compatibility. If you have a 2022 or newer GM car with the Google built-in icon on the infotainment system, you’re in line for Gemini. Make sure your OnStar account is active and you’re signed into the Google Play Store.
- Opt into Gemini when prompted. The update will arrive as a push notification. You’ll need to accept the terms to unlock Gemini. Test the new features—the more natural you speak, the better it works.
- Review your data sharing settings now. Visit your OnStar privacy dashboard and see what information GM is collecting. If you’re uncomfortable with the assistant having access to certain data, disable those categories early. Note that some features, like stolen-vehicle assistance, require data sharing.
- Stay informed about subscription plans. GM hasn’t announced pricing for the integrated assistant. It could be bundled with OnStar plans that already cost $25–$50 per month. If you’re in the market for a new car, ask the dealer about the future AI roadmap and whether it will affect resale value.
- For fleet managers: start auditing your vehicle data policies. If you plan to use the integrated assistant, establish rules about who can query vehicle data and how logs will be stored. Watch for GM to release an enterprise API for the assistant, which would allow integration with fleet management software.
Where GM’s AI Bet Is Heading
GM’s dual-AI rollout is more than a flashy feature; it’s a test of whether car owners will trust an AI with the keys to their vehicle’s brain—and whether automakers can monetize that trust responsibly. As the industry moves toward software-defined vehicles, the car is becoming a platform, and the assistant is becoming the primary interface.
For Windows users who’ve watched Microsoft integrate Copilot into everything from Office to the Edge browser, the pattern is familiar: first you bolt on a general-purpose AI, then you build a domain-specific one that leverages your proprietary data. The success of Copilot in Teams and Outlook depended on its access to Microsoft Graph, just as GM’s assistant will depend on OnStar’s data graph.
The next 12 months will bring answers to critical questions: Can the assistant handle commands safely while driving? Will it remain functional without an active subscription? How will GM secure the connection between the LLM and vehicle controls? And perhaps most importantly, will the assistant actually make ownership easier, or just add another screen to manage?
One thing is certain: by the end of 2026, the phrase “Hey Google” will mean something very different inside a GM car than it does in your living room.