GM is rolling out Google Gemini to roughly 4 million U.S. vehicles already on the road, replacing Google Assistant in model year 2022 and newer Cadillac, Chevrolet, Buick, and GMC models with Google built-in. The update arrives over the air over several months, U.S. English only at launch, with other markets and languages to follow. GM calls it one of the largest Gemini deployments in the auto industry.
The upgrade replaces simple voice commands with conversational AI that maintains context. Drivers can ask for music that matches their mood or trip length, discuss landmarks they’re passing, or explore topics without repeating background information. The system runs through GM’s infotainment platform using OnStar connectivity. GM gets enterprise-grade AI without building it internally. Google gets automotive data and usage patterns that Apple CarPlay and Android Auto cannot provide, since both operate as phone extensions and never reach the vehicle itself.
GM also promises GM Forward: a custom AI assistant fine-tuned with proprietary vehicle data. It arrives later this year. That creates an interesting dependency question. Building custom automotive AI on Google’s foundation could accelerate development. It also means GM’s future AI capabilities depend on continued Google partnership and data sharing agreements.
Then, there is the FTC problem. In January 2025, the agency ordered GM and OnStar to stop selling precise geolocation and driving data to insurers without explicit consent, for five years. Now, GM is layering a large language model that learns driver preferences on top of the same OnStar pipeline. The plumbing has not changed. Have the consent practices? The defaults will tell us.
Voice interfaces in cars have historically disappointed because driving contexts require precision. We’ll see if AI changes that.
Every company needs a Claw strategy. Do you have one?
Author’s note: This is not a sponsored post. I am the author of this article and it expresses my own opinions. I am not, nor is my company, receiving compensation for it. This work was created with the assistance of various generative AI models.