GM is building its own native car AI to move beyond Google Gemini

摘要:General Motors will launch its own in-vehicle AI assistant later this year, aiming to go beyond the Google Gemini system it began rolling out in April to about four million vehicles. The proprietary assistant combines conversational AI with GM vehicle knowledge and OnStar intelligence, enabling predictive maintenance, telemetry monitoring, and a “kids mode” that adjusts locks, seats, temperature, and music. GM has not disclosed the large language model provider or a product name. OnStar telematics give the system its edge, though that data also drew a five-year Federal Trade Commission restriction on sharing driver behavior with insurers. GM is pursuing both tracks: shipping Gemini while building its own assistant, supported by Nvidia-based hardware and a centralized computing platform. The company cited 20% growth in digital services revenue and raised its full-year profit outlook.

General Motors will launch its own in-vehicle artificial intelligence assistant later this year, designed to reach further into the car than the Google Gemini system it started deploying in the spring, per CNBC.

Anna Santos, GMs director of product management of voice and AI/machine learning, said the assistant “combines conversational AI with GM vehicle knowledge and OnStar intelligence” to do things a general-purpose system cannot. She declined to name the product.

The technology is being developed by GM using a large language model provider whose identity they have not revealed. The features under development include predictive maintenance capabilities, telemetry monitoring and a “kids mode” that will lock and unlock doors, change seats, temperature and tunes in one go.

Gemini gives GM a fast AI rollout

GM began the Gemini rollout on April 28 to model year 2022 and newer Cadillac, Chevrolet, Buick, and GMC vehicles with Google built-in, roughly 4 million cars in the United States. Drivers can speak naturally without memorizing commands, and GM has started adding “live sessions” in which the assistant holds looser conversation or plays trivia and 20 questions. It also controls temperature and radio.

Beyond that, CNBC reports, it generally operates the way it would through a phone. That is the gap the native assistant is meant to close, and GM signaled it was coming at its GM Forward event in October 2025 without saying what it would do.

As reported by Cryptopolitan at the time, GM positioned Gemini as the first step towards a dedicated assistant that would work via OnStar.

OnStar data gives GMs assistant the real edge

What makes the proprietary system proprietary is OnStar telematics, and that is the same asset that drew federal action. The Federal Trade Commission barred GM from sharing OnStar Smart Driver data with insurers and consumer reporting agencies without explicit consent for five years, following reporting that driver behavior data reached insurance companies.

GM says drivers control what Gemini can access. The native assistant sharpens the question rather than settling it, because its entire advantage rests on how much of the vehicle and driver it can see. OnStar reaches vehicles built as far back as 2015, and Google has handled non-emergency OnStar queries through Cloud Dialogflow since 2023.

Automakers are choosing different AI partners

Every major automaker has chosen a partner rather than a build. Stellantis is working with French firm Mistral, Mercedes integrated ChatGPT, and Tesla brought in xAI‘s Grok. GM is running both tracks at once, shipping Google’s model to millions of cars while building the system that will sit beneath it.

The business case showed up in the second quarter, when GM reported $48.03 billion in revenue, raised its full-year profit outlook for the second time this year, and posted 20% growth in digital services revenue, per Quartz.

The assistant will also run on hardware GM has been assembling for years, including an Nvidia partnership covering Drive AGX and Omniverse and a centralized computing platform the company says delivers up to 35 times current AI performance, arriving with the Cadillac Escalade IQ in 2028 alongside eyes-off highway driving.

免責聲明

本文觀點僅代表作者個人觀點,不構成本平台的投資建議,本平台不對文章信息準確性、完整性和及時性作出任何保證,亦不對因使用或信賴文章信息引發的任何損失承擔責任
上一篇

劍橋大學研究顯示以太坊(ETH)位居PoS能耗強度下游

下一篇

Merck與Hashgraph Group推出基於Hedera的產品護照以符合歐盟合規要求