GM introduces Google Gemini to four million cars.

      The over-the-air update introduces Google Gemini to Cadillac, Chevrolet, Buick, and GMC vehicles from model year 2022 and newer, but it comes amid ongoing issues regarding GM's data-sharing practices and a forthcoming FTC consent order. General Motors has reported that Google Gemini is being rolled out to around four million vehicles in the U.S., marking what the company describes as one of the largest deployments of a generative AI assistant in the automotive sector.

      This update, announced on April 28 and distributed through an over-the-air Play Store update, will replace the current Google Assistant in eligible Cadillac, Chevrolet, Buick, and GMC vehicles equipped with Google Built-in. Tim Twerdahl, Global Vice President of Product Management at General Motors, stated, “Gemini brings conversational AI to millions of drivers across various segments and price ranges for numerous everyday needs. This level of scale is achievable thanks to the connected vehicle foundation that GM has developed through OnStar over the past 30 years.”

      “Later this year, GM plans to provide a more integrated AI experience influenced by OnStar intelligence,” he added. The claim of scale is credible, with the four million eligible vehicles likely surpassing any current single-OEM deployment of a conversational AI assistant in production vehicles.

      This extensive reach is a direct result of GM's ten-year commitment to Android Automotive OS, the 'Google Built-in' platform that allows Buick, Chevrolet, Cadillac, and GMC vehicles direct access to Google's apps and services, along with the connectivity framework provided by OnStar, GM's in-car connectivity system since 1996.

      The transition from Google Assistant to Gemini represents a shift in conversational capability. The current in-car version of Google Assistant functions primarily as a command-recognition system, effectively executing tasks when drivers use recognized phrases, but it falters with unrecognized language. In contrast, Gemini is a large language model capable of handling open-ended requests, maintaining context throughout a dialogue, responding to follow-up questions without needing to restart the conversation, and showing greater resilience against accent variations and unconventional phrasing.

      For drivers, the most noticeable change will be in the assistant's ability to manage multi-part requests and switch tasks during a conversation. GM's press release exemplifies this: asking for directions while simultaneously texting a family member, then adjusting the route to include a coffee stop with outdoor seating, all within one spoken interaction.

      The assistant connects with in-vehicle applications like Amazon Music, Apple Music, Spotify, YouTube, HBO Max, Hulu, and Prime Video, and can utilize web searches for location and context-aware inquiries. To receive the update, drivers need to be connected to OnStar, signed into the Google Play Store on their infotainment system, and using US English as their language setting. The rollout will occur over several months and will initially be limited to the U.S., with plans for additional markets and languages later.

      For models from 2025 and onwards, access to basic OnStar voice features, and thus to Gemini, is included in the standard OnStar Basics package for a duration of eight years at no extra cost. GM has made it clear that Gemini is a temporary solution. The company aims to deploy a custom-built AI assistant, fine-tuned using proprietary vehicle data and linked through OnStar, which will effectively be a domain-specific model capable of understanding every detail of an individual vehicle, identifying maintenance issues before they escalate, and learning personal preferences over time. This advanced assistant is expected to arrive ‘later this year.’

      Gemini serves as a commercial intermediate: it provides GM with four million users experiencing a significantly improved in-car AI now while the company continues to develop the vehicle-specific capabilities. Dave Richardson, GM's SVP of Software and Services, described the architectural approach as starting with a base model, refining it with vehicle specifications, and implementing it directly in the vehicle.

      This hybrid architecture, combining on-vehicle and cloud resources, will be crucial as the number of models increases, regulatory scrutiny on connected vehicle data tightens, and connectivity varies by market. The competitive landscape is bustling and rapidly evolving. Stellantis is collaborating with French AI firm Mistral on in-car assistants, Mercedes-Benz has integrated ChatGPT, and Tesla has introduced xAI’s Grok across its fleet. BMW is also pursuing its own AI assistant program. GM's strategy is more gradual than Tesla's all-in-one approach; it is leveraging Android Automotive and Gemini while layering its own technology on top. However, the four-million-vehicle deployment scale distinguishes GM from its competitors in a significant way.

      This announcement coincides with a controversial data issue. In January 2025, the Federal Trade Commission acted against GM and OnStar regarding their collection and sale of precise geolocation and driving behavior data to insurance companies, supposedly without adequate consumer consent. This consent order prohibits GM from selling such data without explicit permission for five years. Concerns surrounding GM's data practices, including reports of shared driving scores with insurers leading to premium spikes for unsuspecting drivers, attracted considerable