What AI platform does General Motors use?

General Motors has named no company-wide employee AI platform; its AI runs as systems it builds in-house, including the OnStar AI Assistant, the PlantBrAIn factory platform, an AI virtual wind tunnel and the Road to Sim autonomy pipeline (Oct 2026).

Which AI models does General Motors use?

General Motors uses Google Gemini in cars with Google built-in, works with an unnamed large language model provider on its own OnStar AI Assistant (Jul 2026), and builds its own vision-language-action and world models for autonomy (Mar 2026).

How many GM employees use AI?

General Motors' AI reaches customers first: about 4 million U.S. vehicles were eligible for Gemini in Apr 2026. Inside GM, Mary Barra's senior leaders compare AI usage by function and department every month, but GM has published no employee user count (Sep 2026).

How far has AI spread at General Motors?

General Motors sits at Piloting: AI runs in production in its vehicles (Gemini for about 4 million eligible U.S. vehicles, Super Cruise), its plants (machine learning quality checks and the PlantBrAIn platform) and its engineering (an AI virtual wind tunnel and AI design tools), and Mary Barra said in Oct 2026 that every function uses AI. General Motors falls short of Scaling on three tests: it has published no company-wide employee AI platform or user count, it has named no successor to Chief AI Officer Barak Turovsky, who left in Nov 2025, and it reports no AI results, with CFO Paul Jacobson calling its AI ambitions in their infancy in Sep 2026.

Verified Nov 1, 2026

General Motors

Piloting

Automotive
~155,000 employees (2025)
~4M U.S. vehicles eligible for Gemini
OnStar AI Assistant due
~600 IT jobs cut (2026)

General Motors deploys AI mainly in its vehicles and engineering: Google Gemini began rolling out to about 4 million eligible U.S. vehicles in 2026, a GM-built OnStar AI Assistant is due on the 2027 Cadillac Lyriq, and an in-house autonomy stack trains on fleet data and simulation for eyes-off driving in 2028. Inside the company, General Motors tracks AI use monthly by function, applies AI to design, virtual testing and plant quality, and cut about 600 IT jobs in May 2026 while hiring for AI skills.

gm.com

Their approach to AI

The platform: General Motors has named no company-wide employee AI platform; its AI runs as systems it builds in-house, including the OnStar AI Assistant, the PlantBrAIn factory platform, an AI virtual wind tunnel and the Road to Sim autonomy pipeline (Oct 2026).
The models: General Motors uses Google Gemini in cars with Google built-in, works with an unnamed large language model provider on its own OnStar AI Assistant (Jul 2026), and builds its own vision-language-action and world models for autonomy (Mar 2026).
The reach: General Motors' AI reaches customers first: about 4 million U.S. vehicles were eligible for Gemini in Apr 2026. Inside GM, Mary Barra's senior leaders compare AI usage by function and department every month, but GM has published no employee user count (Sep 2026).
The catch: General Motors has credited no dollar savings to AI: Mary Barra said on Oct 20, 2026 that every function uses AI and declined to quantify the fixed-cost savings, and CFO Paul Jacobson called GM's AI ambitions in their infancy in Sep 2026.

Snapshot

Employees: ~155,000; Dec 31, 2025, worldwide; about 47,000 U.S. salaried (2025 10-K)
Technology budget: Not disclosed; R&D expense $8.5B in 2025; $1B to $1.5B in 2026 for onshoring and software, split about evenly
AI/ML headcount: Not disclosed; 100+ roboticists, AI engineers and hardware specialists in GM's robotics labs (Oct 2025)
Who owns AI: Sterling Anderson, CPO; Owns product, software and autonomy; no successor named to Chief AI Officer Barak Turovsky, who left in Nov 2025
Platform: No named employee platform; OnStar AI Assistant, PlantBrAIn and the autonomy stack built in-house
Models: Google Gemini, undisclosed LLM, in-house; Own vision-language-action and world models for autonomy (Mar 2026)
Reach: ~4M vehicles eligible for Gemini; Apr 2026; usage reported monthly by function, no employee count published
Use cases: AI in every function; Mary Barra, Oct 20, 2026; no count published

Key findings

Google Gemini began rolling out in 2026 to about 4 million eligible model year 2022 and newer Cadillac, Chevrolet, Buick and GMC vehicles in the U.S., and General Motors said on Oct 21, 2026 that its own OnStar AI Assistant will reach the 2027 Cadillac Lyriq by over-the-air update in the coming months.
General Motors named Barak Turovsky its first Chief AI Officer in Mar 2025, reporting to software chief Dave Richardson; both left in late 2025 as Sterling Anderson consolidated GM's technology teams, the AI team moved under manufacturing engineering, and in May 2026 GM cut about 600 IT jobs, more than 10% of IT, while hiring for AI skills, TechCrunch reported.
CFO Paul Jacobson said on Sep 23, 2026 that Mary Barra's senior leaders compare AI usage by function and department every month and called General Motors' AI ambitions in their infancy; Barra said on Oct 20, 2026 that every function uses AI and that GM could quantify fixed-cost savings with its 2027 guidance.
General Motors aims to halve its four-to-five-year vehicle development cycle to about two years with enterprise AI and virtualization, and said in Oct 2026 that testing more than 300,000 LMR battery designs virtually took the team from first pack concept to program approval in one month.
General Motors' autonomy stack drives more than 3 billion virtual miles a day in simulation and learns from about 1 million miles a month of fleet sensor data, aimed at eyes-off highway driving on the Cadillac Escalade IQ in 2028 (Oct 2026).

Where GM sits

Stage: Piloting
General Motors sits at Piloting: AI runs in production in its vehicles (Gemini for about 4 million eligible U.S. vehicles, Super Cruise), its plants (machine learning quality checks and the PlantBrAIn platform) and its engineering (an AI virtual wind tunnel and AI design tools), and Mary Barra said in Oct 2026 that every function uses AI. General Motors falls short of Scaling on three tests: it has published no company-wide employee AI platform or user count, it has named no successor to Chief AI Officer Barak Turovsky, who left in Nov 2025, and it reports no AI results, with CFO Paul Jacobson calling its AI ambitions in their infancy in Sep 2026.
How the stages are defined
Ownership: CPO Sterling Anderson for product AI; AI team under manufacturing engineering since Nov 2025
Platform: Hybrid: Autonomy, OnStar AI and factory AI built in-house; Gemini bought from Google
Models: Google Gemini in cars; unnamed LLM under OnStar AI; own autonomy models
Hosting: Vehicle compute and unnamed clouds: Google runs Gemini and OnStar voice AI; NVIDIA GPUs for training; Azure named preferred cloud in 2021
Reach: ~4M U.S. vehicles eligible for Gemini (Apr 2026); employee use not disclosed

How they deploy

AI in the product first

General Motors describes a deeply integrated AI strategy that combines shared compute platforms, proprietary data pipelines and specialized models (GM Forward, Oct 21, 2026). Most of what GM has shipped faces customers: Gemini in the car, OnStar rebuilt as a multimodal intelligence platform for drivers and fleets, Super Cruise, and from 2028 a central compute unit with about 40 times the computing power of earlier systems to run in-cabin AI and automated driving together.

Buy the general model, build the vehicle-specific one

General Motors chose Google's Gemini for general conversation and is building its own OnStar AI Assistant on vehicle knowledge and OnStar data. Dave Richardson, then senior vice president of software and services, said in Oct 2025 that GM would take a base large language model, train it on vehicle specifications, distill it to run on the vehicle, and test foundation models from other firms, possibly including OpenAI and Anthropic. CNBC reported in Jul 2026 that GM is working with an unnamed large language model provider on its assistant, and TechCrunch described the OnStar AI Assistant in Oct 2026 as an in-house effort.

Halve the vehicle development cycle

General Motors is working to cut its four-to-five-year vehicle development timeline in half through enterprise AI and virtualization, with design, engineering and manufacturing working from one digital thread (Oct 2026). Jason Fischer, executive director of virtual integration engineering, told Business Insider in Jun 2026 that GM rarely uses an exact off-the-shelf tool, customizes tools with software suppliers, builds some itself and owns the IP on several techniques. GM declined to say what it budgets for AI use by designers and engineers and said it watches whether AI solves business problems more than token volume.

Autonomy rebuilt in-house after Cruise

General Motors stopped funding Cruise's robotaxi work in Dec 2024, wound down its robotaxi operations in 2025 and combined Cruise's personal autonomy work with its own (2025 10-K). GM now builds one generalized Compound AI driving stack, an end-to-end learned system with perception outputs and guardrails, and distills every product from entry driver assistance to eyes-off driving from it (Mar 2026). GM is hiring from Tesla, Waymo and Zoox, and a spokesperson said in Jun 2026 that GM had nearly doubled the prior year's external autonomy hires.

Usage tracked, savings promised

General Motors' senior leaders compare AI usage by function and department every month, CFO Paul Jacobson said in Sep 2026, and Jacobson wants AI to flip finance analysts from spending 80% of their time explaining past results to 80% predicting what comes next. Mary Barra said on Oct 20, 2026 that GM is working to make sure everyone uses AI to make their jobs, design and manufacturing more efficient, and to turn that into fixed-cost savings, which she said GM could quantify with its 2027 guidance.

Deployment model: General Motors builds most of its AI inside product, engineering and plant teams rather than in a central AI office. The Global Product organization under Sterling Anderson, executive vice president and chief product officer since Jun 2025, who reports to President Mark Reuss, owns hardware, software, services, autonomy and OnStar, and the 2025 10-K describes it as combining software, AI and vehicle platforms. Anderson took over the combined vehicle software engineering and global product teams in late 2025, and GM's first Chief AI Officer, Barak Turovsky, who sat in software and services engineering, left in Nov 2025, when GM said the AI team would report to the manufacturing engineering organization (CIO Dive). Behrad Toghi, described by TechCrunch as GM's AI lead, is listed as head of AI robotics engineering at GM's Mountain View Tech Center. GM's IT organization is being rebuilt around AI skills (May 2026), and senior leaders reporting to Mary Barra review AI usage by function every month (Sep 2026).
Model providers: General Motors uses Google's Gemini as the conversational assistant in vehicles with Google built-in, an upgrade from Google Assistant that began rolling out in 2026, and OnStar's virtual assistant has used Google Cloud conversational AI since 2022, moving to Google Cloud's Dialogflow in 2023. GM is building its own OnStar AI Assistant, which combines conversational AI with GM vehicle knowledge, with an unnamed large language model provider (CNBC, Jul 2026). For autonomy, General Motors builds its own models, including vision-language-action models fine-tuned from a standard vision-language model and a GM World Model that generates synthetic driving scenes (Mar 2026), and its engineers have tested open-source large language models on embedded C code (2026 SAE paper cited by GM). Design teams use tools built with the AI startup Vizcom (Mar 2026).
Data and infrastructure: General Motors has invested in NVIDIA GPU platforms for AI model training and in Mar 2025 extended the partnership to NVIDIA Omniverse digital twins of assembly lines and NVIDIA DRIVE AGX in-vehicle hardware for future driver assistance. GM's autonomy data comes from hundreds of data-collection vehicles adding about 1 million miles of multisensor data a month and from retail vehicles whose owners opted in, and its Road to Sim pipeline fans out across thousands of cloud workers on a cloud GM does not name (Oct 2026). In Jan 2021 General Motors named Microsoft Azure its preferred public cloud for digitization, including AI and machine learning, and has not restated that choice since. The OnStar AI Assistant runs in the vehicle and, from 2027, in a new OnStar phone app; GM has not said where its model inference runs. From 2028 a GM-designed, liquid-cooled central compute unit will run in-cabin AI and automated driving on the vehicle.

Platforms and internal tools

OnStar AI AssistantAnnounced Oct 2026; debuts on the 2027 Cadillac Lyriq
OnStar AI Assistant is General Motors' own in-vehicle assistant, built on the driver's vehicle, context and GM knowledge, that adjusts vehicle functions by voice, reports tire pressure, range and charging status, explains features for the exact model, and hands off to a human OnStar Advisor when needed. GM said on Oct 21, 2026 that it will reach the 2027 Cadillac Lyriq by over-the-air update in the coming months, other GM vehicles later, and a new OnStar app from 2027.
Models
Undisclosed large language model provider plus GM vehicle data
Source
Google Gemini in GM vehicles2026
Google Gemini replaced Google Assistant as the conversational assistant in model year 2022 and newer Cadillac, Chevrolet, Buick and GMC vehicles with Google built-in, delivered over the air over several months; Tim Twerdahl said the scale depends on the connected-vehicle base GM built through OnStar.
Users
About 4 million eligible U.S. vehicles (Apr 2026)
Models
Google Gemini
Source
OnStar Fleet Intelligence and Fleet AI AssistantSep 2026
OnStar Fleet Intelligence brings fleet vehicle data and connected services into one dashboard, and its built-in Fleet AI Assistant answers fleet managers' natural-language questions, such as idling time per vehicle, to cut manual analysis and reporting.
Source
PlantBrAInBy Oct 2026
PlantBrAIn is General Motors' factory intelligence platform, which combines plant-floor data with computer models to improve how vehicles and battery packs are built; GM will use it for fourth-generation EVs and says similar tools run across stamping, casting, welding, painting and final assembly.
Source
WeldBrAIn and Device Level AnalyticsPilot by May 2026
WeldBrAIn is a proprietary AI system GM is piloting at Factory ZERO that checks every weld on every body in real time, where manual inspection covered four parts per shift. Device Level Analytics uses plant sensors that capture vibration and electrical current to schedule preventive maintenance, and GM plans to add machine learning models so it flags issues faster.
Source
AI virtual wind tunnelBy Mar 2026
GM R&D built an AI virtual wind tunnel, trained on GM's own computational fluid dynamics data, that predicts aerodynamic drag inside the digital sculpting tools designers already use, returning results in about a minute where a design and engineering cycle took about two weeks.
Source
Road to Sim and the autonomy simulation stackBy Oct 2026
Road to Sim converts road logs into simulation scenarios, runs them as regression tests in continuous integration across thousands of cloud workers, and uses AI agents with vision-language models to grade conversion quality alongside human reviewers. GM's simulation drives more than 3 billion virtual miles a day, and its Boxworld reinforcement learning environment runs 50,000 times faster than real time (Mar 2026).
Source
EMWU fleet data miningBy Jun 2026
EMWU is an offboard pipeline that mines historical fleet video for rare driving scenarios, running cheap detection and embeddings first and saving vision-language model reasoning for the few candidate clips that need it; GM says it produces tens of millions of embeddings a day at under $1.00 per million.
Source
Autonomous Robotics CenterBy Oct 2025
The Autonomous Robotics Center in Warren, Michigan, with a sister lab in Mountain View, California, had more than 100 roboticists, AI engineers and hardware specialists in Oct 2025 building robotics systems trained on GM production data, and develops software for the collaborative robots GM began deploying in U.S. assembly plants in 2025.
Source

Vendors and partners

Google2019
Gemini in GM vehicles; Google built-in; Google Cloud conversational AI for OnStar · GM has built vehicles with Google built-in since 2019, OnStar's virtual assistant has used Google Cloud conversational AI since 2022, Gemini began rolling out to about 4 million eligible U.S. vehicles in 2026, and GM began a gradual Portuguese-language Gemini rollout to Chevrolet vehicles in Brazil on Oct 28, 2026.
NVIDIAExpanded Mar 2025
GPUs for AI model training, Omniverse digital twins of assembly lines, DRIVE AGX in-vehicle hardware · General Motors said in Mar 2025 that it had already been investing in NVIDIA GPU platforms for AI model training before the expanded collaboration.
MicrosoftJan 2021
Named GM's preferred public cloud, including AI and machine learning · The Jan 2021 agreement also made Azure Cruise's preferred cloud; GM has not said whether the arrangement continued after it wound down Cruise's robotaxi operations.
Undisclosed large language model providerBy Jul 2026
Model technology under the OnStar AI Assistant · CNBC reported in Jul 2026 that GM is working with an unnamed large language model provider on its own assistant; Dave Richardson had said in Oct 2025 that GM would test models from several firms.
VizcomBy Mar 2026
AI design startup; tools that turn hand sketches into 360-degree models and animations
Mitra Chem
AI-enabled battery materials developer · General Motors led a $60 million investment in Mitra Chem, which simulates, synthesizes and tests thousands of cathode designs a month, according to GM's AI page.

Leadership

AI, data and platform leaders

Sterling Anderson

Executive Vice President, Global Product, and Chief Product Officer

Sterling Anderson joined General Motors on Jun 2, 2025 from Aurora Innovation, which he co-founded in 2017; before that he led the Model X program and the team that delivered Autopilot at Tesla, worked at McKinsey, and earned a PhD in robotics at MIT. Anderson leads a product organization of more than 20,000 people (Oct 2025) covering hardware, software, services and user experience, and led GM Forward in Oct 2026, where GM set out its AI plans.

Rashed Haq

Vice President, Autonomy

Rashed Haq joined General Motors in 2025, having earlier spent five years as head of AI and robotics at Cruise, and runs the autonomy organization building eyes-off driving for the 2028 Cadillac Escalade IQ; GM's engineering blog called him Vice President of Autonomy in Oct 2026. Haq told Business Insider in Jun 2026 that GM's new autonomy stack ran in simulation in Jan 2026, on a closed course in Feb and on public roads in Mar, and that GM hires engineers from Tesla, Waymo and Zoox.

Christina Montgomery

Chief Privacy & Trust Officer and Assistant General Counsel

Unconfirmed

Christina Montgomery joined General Motors in 2025 from IBM, where she was Vice President and Chief Privacy & Responsible Technology Officer and created and co-chaired IBM's AI Ethics Board; her NACD speaker biography says her GM remit covers privacy strategy and compliance, including data and AI governance. TechCrunch named Montgomery in Oct 2025 as part of a new GM data team, and GM has not published her role on its own site.

Behrad Toghi

AI lead; Head of AI Robotics, Engineering, Mountain View Tech Center

Unconfirmed

Behrad Toghi, who previously worked at Apple, is General Motors' AI lead according to TechCrunch, which dated the appointment to Oct 2025, and an SAE International webinar listing gives his title as Head of AI Robotics, Engineering, at GM's Mountain View Tech Center. GM has not published his remit or reporting line.

Tim Twerdahl

Global Vice President, Product Management

Tim Twerdahl announced the Gemini rollout to about 4 million eligible U.S. vehicles in Apr 2026 and said General Motors would follow it with a more deeply integrated AI experience shaped by OnStar.

Mike Hanley

Vice President and Chief Information Security Officer

Mike Hanley is General Motors' CISO, named in an Aug 2026 GM News story. GM's 2025 10-K says its CISO has served since Dec 2024, has more than 20 years in IT, cybersecurity and software engineering roles, and reports to the board's Risk and Cybersecurity Committee.

Executives

Mary Barra

Chair and Chief Executive Officer

Mary Barra has been CEO of General Motors since 2014 and Chair since 2016, and joined GM in 1980 as a co-op student; she earlier ran Global Product Development, Purchasing and Supply Chain. Barra said on Oct 20, 2026 that every GM function uses AI and that GM will look to turn AI use into fixed-cost savings, which she declined to quantify before the company's 2027 guidance.

Mark Reuss

President

Mark Reuss has been President of General Motors since Jan 2019 and oversees its product and operating organizations, including Global Product, Quality and Manufacturing. Reuss earlier ran Global Product Development, Purchasing and Supply Chain and GM North America, and Chief Product Officer Sterling Anderson reports to him.

Paul Jacobson

Executive Vice President and Chief Financial Officer

Paul Jacobson joined General Motors as CFO in Dec 2020 from Delta Air Lines, where he was CFO. Jacobson said in Sep 2026 that GM's senior leaders compare AI usage by function every month, that his own team was in the middle of the pack, and that GM's AI ambitions are in their infancy.

Grant Dixton

Executive Vice President, Chief Legal and Public Policy Officer, and Corporate Secretary

Grant Dixton joined General Motors in Jul 2024 from Activision Blizzard, where he was chief legal officer, after executive roles at Boeing that included general counsel. Dixton reports to Mary Barra and oversees GM's global legal, compliance, corporate governance, privacy and public policy functions, the area where Christina Montgomery leads privacy and AI governance; GM has not published her reporting line.

Board

Devin Wenig

Director; chair of the Compensation Committee

Devin Wenig has been a General Motors director since 2018 and since 2023 has been co-founder and CEO of Symbolic.ai, an AI platform for publishers and writers; he was earlier President and CEO of eBay. GM's 2026 proxy cites his experience in software and technology, including AI, and asked shareholders, on the board's recommendation, for more equity to recruit software, robotics and AI talent; shareholders re-elected Wenig and approved the shares on Jun 2, 2026.

AI org map

AI built inside product, engineering and plant teams; no central AI office named since Nov 2025

General Motors created a Chief AI Officer role in Mar 2025 for Barak Turovsky inside software and services engineering under Dave Richardson. Richardson stepped down at the end of Oct 2025 and Turovsky and Baris Cetinok, senior vice president of software and services product, also left as Sterling Anderson consolidated GM's technology businesses into one organization; GM said the AI team would report to manufacturing engineering. Product AI, software, autonomy and OnStar now sit under Chief Product Officer Sterling Anderson, who reports to President Mark Reuss.

General Motors has not published how many people work on AI. The Global Product organization under Sterling Anderson had more than 20,000 people in Oct 2025, GM's robotics center and its Mountain View sister lab had more than 100 roboticists, AI engineers and hardware specialists, and a GM spokesperson said in Jun 2026 that external autonomy hires had nearly doubled from 2025. AI work also sits in the autonomy organization, where Ben Snyder wrote as Director of AI Research in Mar 2026, and in GM R&D, design, virtual integration engineering and plant teams; the former Chief AI Officer's team reports to manufacturing engineering (Nov 2025), and after General Motors cut about 600 IT jobs in May 2026, TechCrunch reported that GM is hiring IT staff for AI-native development, data engineering, cloud engineering and agent and model development.

Reporting lines

Chair and CEO
Mary Barra
- President
  Mark Reuss
  - Global Product
    Sterling Anderson, EVP and Chief Product Officer
    Hardware, software, services and user experience for gas and electric vehicles; more than 20,000 people (Oct 2025)
    - Autonomy
      Rashed Haq, VP
      Eyes-off driving for 2028; the line to Global Product is inferred, not published
    - Product Management
      Tim Twerdahl, Global VP
      Gemini and the OnStar AI Assistant; the line to Global Product is inferred, not published
    - Virtual Integration Engineering
      Jason Fischer, Executive Director
      Virtual testing and co-simulation; the line to Global Product is inferred, not published
  - AI team under manufacturing engineering
    The former Chief AI Officer's team, moved under the manufacturing engineering organization in Nov 2025 (CIO Dive); Mark Reuss oversees Manufacturing, so the line to the President is inferred, and the team's leader is not published
- Chief Financial Officer
  Paul Jacobson
- Legal, privacy and public policy
  Grant Dixton, EVP and Chief Legal and Public Policy Officer
  Global legal, compliance, corporate governance, privacy and public policy (Jul 2024 announcement)
  - Privacy and AI governance
    Christina Montgomery, Chief Privacy & Trust Officer
    Data and AI governance, per her NACD speaker biography; the line to Legal is inferred from her Assistant General Counsel title, not published
AI Robotics, Mountain View Tech Center
Behrad Toghi, Head of AI Robotics
Described by TechCrunch as GM's AI lead (May 2026); reporting line not published
Information Technology
Restructured in May 2026, cutting about 600 salaried jobs and hiring for AI skills; leader not published
Autonomous Robotics Center
Warren, Michigan, with a sister lab in Mountain View; reporting line not published
Cybersecurity
Mike Hanley, VP and CISO
Reports on cyber risk to the board's Risk and Cybersecurity Committee (2025 10-K); management reporting line not published

General Motors' officer biographies, checked Nov 1, 2026, say Mark Reuss oversees Global Product, Quality and Manufacturing, and the May 2025 announcement says Sterling Anderson reports to Reuss; the Jul 2024 announcement says Grant Dixton reports to Mary Barra, and the other lines come from the 2025 10-K list of executive officers. GM has not published who leads its AI team after Nov 2025, who runs Information Technology, or the reporting lines of Behrad Toghi, Rashed Haq, Tim Twerdahl, Jason Fischer, Christina Montgomery and Mike Hanley.

Where the AI workforce sits

Global Product: software, autonomy, OnStar and virtual engineering
General Motors has not published AI headcount; the autonomy organization is, in Rashed Haq's words, appropriately sized (Jun 2026).
Not disclosed
Manufacturing engineering and robotics
More than 100 roboticists, AI engineers and hardware specialists in GM's robotics labs (Oct 2025), plus the former Chief AI Officer's team.
Not disclosed
GM R&D and design
GM R&D built the AI virtual wind tunnel, and design runs AI visualization tools with Vizcom.
Not disclosed

Named people

Executive sponsors

Mary Barra
Chair and CEO
Mark Reuss
President
Paul Jacobson
EVP and Chief Financial Officer · Reports monthly AI usage competition among senior leaders (Sep 2026)

Product, software and autonomy

More than 20,000 in Global Product (Oct 2025)

Sterling Anderson
EVP, Global Product, and Chief Product Officer
Rashed Haq
Vice President, Autonomy
Ben Snyder
Director of AI Research · Wrote GM's Mar 2026 engineering post on training driving AI at 50,000 times real time
Tim Twerdahl
Global Vice President, Product Management
Anna Santos
Director, product management, voice and AI/machine learning · Described the OnStar AI Assistant plans to CNBC (Jul 2026)

Design, virtual engineering and R&D

Bryan Styles
Director, Design Innovation and Technology Operations
Jason Fischer
Executive Director, Virtual Integration Engineering
Rene Strauss
Director, Virtual Integration Engineering
Scott Parrish
Technical Fellow and Lab Group Manager, GM R&D · His team built the AI virtual wind tunnel

Manufacturing and robotics

More than 100 roboticists, AI engineers and hardware specialists (Oct 2025)

GM said in Nov 2025 that the AI team would report to the manufacturing engineering organization (CIO Dive).

Behrad Toghi
AI lead; Head of AI Robotics, Engineering, Mountain View Tech Center
Mikell Taylor
Director of Robotics Strategy, Autonomous Robotics Center

Security, privacy and AI governance

Mike Hanley
Vice President and Chief Information Security Officer
Christina Montgomery
Chief Privacy & Trust Officer and Assistant General Counsel · Remit includes data and AI governance, per her NACD speaker biography

Use cases

General Motors' public AI use cases sit in four places: the vehicle and OnStar (Gemini, the OnStar AI Assistant, the fleet assistant, Super Cruise), autonomy development, product engineering (AI design tools, the virtual wind tunnel, virtual crash analysis) and plant quality, and Mary Barra said on Oct 20, 2026 that every GM function uses AI. General Motors has published no business return, such as a cost, revenue or headcount figure, that it credits to AI for any of these use cases; the gains it reports are process speedups, such as faster battery-pack assembly with PlantBrAIn and faster crash and aerodynamic analysis. For AI as a whole General Motors reports no savings figure; Barra said GM could put a number on AI-driven fixed-cost savings when it gives 2027 guidance.

Use case	Area	Status
Google Gemini in-vehicle assistant Gemini answers drivers in natural conversation, drafts and translates messages and plans multi-stop routes in model year 2022 and newer GM vehicles with Google built-in, delivered over the air from 2026; CNBC reported in Jul 2026 that it also controls some vehicle functions, such as temperature and radio.	Vehicles and OnStar	In production
OnStar AI Assistant GM's own assistant, built on vehicle data and GM knowledge, adjusts vehicle functions by voice, explains warning lights and features for the exact model, plans charging stops and connects drivers to an OnStar Advisor; it debuts on the 2027 Cadillac Lyriq.	Vehicles and OnStar	Announced
OnStar virtual assistant for non-emergency calls OnStar's virtual assistant, on Google Cloud conversational AI since 2022, handles routine requests such as routing and vehicle questions, detects words that signal an emergency and routes those calls to human advisors; it handled more than 1 million inquiries a month in the U.S. and Canada in Aug 2023.	Vehicles and OnStar	In production
OnStar Fleet AI Assistant The Fleet AI Assistant inside OnStar Fleet Intelligence answers fleet managers' natural-language questions about idling, utilization and vehicle health to reduce manual analysis and reporting.	Vehicles and OnStar	In production
Super Cruise hands-free driving Super Cruise, the hands-free driver assistance system GM launched in 2017, had logged more than 1.2 billion hands-free miles across nearly 750,000 vehicles by Oct 2026, and GM says data from retail vehicles whose owners opted in feeds its autonomy development.	Vehicles and autonomy	In production
Eyes-off highway driving GM's next-generation autonomy stack, trained on fleet data, simulation, imitation learning and reinforcement learning through self-play, began supervised public-road testing in California and Michigan in Mar 2026 and targets the Cadillac Escalade IQ in 2028, then full-size SUVs and pickups.	Vehicles and autonomy	Pilot
Road to Sim and AI-graded simulation Road to Sim turns road logs into editable simulation scenarios that run as regression tests, and AI agents using vision-language models flag and describe mismatches between road and simulated video for human reviewers.	Autonomy development	In production
Fleet data mining with EMWU EMWU mines millions of fleet video clips for rare scenarios such as near misses and hard braking, using low-cost detection and embeddings first and vision-language model reasoning only on shortlisted clips.	Autonomy development	In production
AI design visualization GM designers feed hand sketches into AI tools, which Business Insider reported GM built with the startup Vizcom, to generate 3D views, variations and teaser animations that a single designer can now produce in less than a day instead of multiple teams over months.	Product development	In production
AI virtual wind tunnel An AI model trained on GM's computational fluid dynamics data predicts drag inside designers' sculpting tools in about a minute, so designers and aerodynamicists can trade off surface changes together.	Product development	In production
Machine learning crash simulation GM uses machine learning and simulation to cut roof crush analysis runs from 8 to 40 hours to under 5 minutes, running more iterations before physical crash tests.	Product development	In production
AI-assisted embedded software engineering GM researchers use large language models for bounded tasks in vehicle software, including test-case generation, refactoring embedded C, translating Perl utilities to Python and an Intelligent Virtual Advisor that flags coding-standard violations, with engineers approving every step.	Product development	Pilot
PlantBrAIn factory intelligence PlantBrAIn combines plant-floor data with computer models to refine how vehicles and battery packs are assembled, and GM will use it to build fourth-generation EVs; GM said in Oct 2026 that the approach has moved battery packs through its pack-assembly process 30% faster.	Manufacturing	In production
Predictive weld quality A cross-functional team built a predictive weld quality model that Mary Barra said gives more consistent welds and tighter control in GM plants (Jan 2026).	Manufacturing	In production
WeldBrAIn weld monitoring WeldBrAIn, piloted at Factory ZERO, checks every weld on every body in real time, where manual inspection covered four parts per shift.	Manufacturing	Pilot
Device Level Analytics and paint vision inspection Plant sensors feed Device Level Analytics to schedule preventive maintenance, with machine learning models being added, and vision cameras at the Spring Hill and Flint paint shops detect microscopic paint defects and feed data to paint engineers.	Manufacturing	In production
Collaborative robots GM's Autonomous Robotics Center develops software and manipulation components for collaborative robots that work beside people in U.S. assembly plants, trained on GM production data.	Manufacturing	In production
Digital twins of assembly lines GM is building digital twins of assembly lines on NVIDIA Omniverse to test production steps, equipment placement and ergonomics virtually before hardware is installed.	Manufacturing	Pilot
Supply chain risk alerts GM told Business Insider that AI tools alert the company to potential risks in its supply chain.	Supply chain	In production
EV charger site selection GM data scientists use predictive analytics and geospatial algorithms on EV traffic patterns to choose charger locations, which GM then builds with Pilot Flying J and EVgo.	EV infrastructure	In production

Adoption

General Motors has used machine learning in its products for years, with Super Cruise from 2017 and an OnStar virtual assistant on Google Cloud conversational AI from 2022, and widened AI into engineering and plants from 2025: a Chief AI Officer in Mar 2025, an expanded NVIDIA partnership the same month, AI design and virtual testing tools by early 2026, and PlantBrAIn in battery-pack assembly by Oct 2026. Customer reach is the largest published number, with about 4 million U.S. vehicles eligible for Gemini in Apr 2026. Employee adoption is tracked but not published: Paul Jacobson said in Sep 2026 that senior leaders compare AI usage by function every month, and Mary Barra said in Oct 2026 that every function is using AI.

By the numbers

U.S. vehicles eligible for Gemini: ~4M; Model year 2022 and newer with Google built-in; announced Apr 2026, delivered over several months

Virtual miles driven by GM's autonomy stack each day: 3B+; Simulation with reinforcement learning through self-play (GM Forward, Oct 21, 2026)

Fleet sensor data collected for autonomy: ~1M miles a month; Hundreds of data-collection vehicles, full-rate multisensor data (Oct 2026)

Research and development expense (2025): $8.5B; Down from $9.2B in 2024; no IT or AI budget published (2025 10-K)

Employees: ~155,000; Dec 31, 2025, about 88,000 hourly and 68,000 salaried (2025 10-K)

Rollout, first pilot to today

2017
General Motors became the first automaker to offer hands-free driving with Super Cruise.
GM News
Jan 2021
General Motors named Microsoft its preferred public cloud for digitization, including AI and machine learning, and Azure became Cruise's preferred cloud.
Microsoft
Aug 2023
OnStar's virtual assistant, on Google Cloud conversational AI since 2022, moved to Google Cloud's Dialogflow and was handling more than 1 million inquiries a month.
TechCrunch
Dec 2024
General Motors stopped funding Cruise's robotaxi work and refocused autonomy on personal vehicles, later folding Cruise's team into GM.
2025 Form 10-K
Mar 2025
General Motors named Barak Turovsky its first Chief AI Officer, reporting to Dave Richardson, senior vice president of software and services engineering.
WardsAuto
Mar 2025
General Motors expanded its NVIDIA partnership from GPUs for model training to Omniverse digital twins of assembly lines and DRIVE AGX in-vehicle hardware.
GM News
Jun 2025
Sterling Anderson joined General Motors as EVP, Global Product, and Chief Product Officer, with responsibility for hardware, software, services and user experience.
GM News
Oct 2025
At GM Forward, General Motors announced Gemini for 2026, its own assistant to follow, eyes-off driving for 2028 and a robotics center with more than 100 roboticists and AI engineers.
GM News
Nov 2025
Chief AI Officer Barak Turovsky resigned, a month after Dave Richardson stepped down, and GM said its AI team would report to the manufacturing engineering organization.
CIO Dive via Yahoo Finance
Mar 2026
General Motors began supervised public-road testing of its next-generation automated driving technology on highways in California and Michigan.
GM News
Apr 2026
General Motors announced that Gemini would roll out over the air, over several months, to about 4 million eligible model year 2022 and newer U.S. vehicles with Google built-in.
GM News
May 2026
General Motors cut about 600 salaried IT jobs, more than 10% of the department, to make room for hires in AI-native development, data, cloud and agent and model development.
TechCrunch
Sep 2026
Paul Jacobson said General Motors' senior leaders compare AI usage by function and department every month.
Business Insider
Oct 2026
General Motors announced the OnStar AI Assistant for the 2027 Cadillac Lyriq and set a goal of halving vehicle development to about two years with enterprise AI and virtualization.
GM News
Oct 2026
Gemini reached Chevrolet vehicles in Brazil in Portuguese, after GM's Spanish-speaking Latin American markets.
AutoPapo

Economics & controls

Spend and ROI

Technology and R&D spending

General Motors does not publish a technology or AI budget. Research and development expense was $8.5 billion in 2025, down from $9.2 billion in 2024 and $9.9 billion in 2023 (2025 10-K), and GM is spending $1 billion to $1.5 billion in 2026 to onshore production, strengthen its supply chain and expand software capabilities, which Paul Jacobson split about evenly between onshoring and software (Jan 2026). GM's 2026 proxy asked shareholders for more equity to recruit talent in software, robotics and AI, which they approved on Jun 2, 2026 by adding 27 million shares to the long-term incentive plan, and describes $24 million in new-hire awards for Sterling Anderson.

What AI itself costs

General Motors declined in Jun 2026 to say how much it budgets for AI use by designers and engineers, saying it watches whether AI solves business problems more than token volume. GM engineers describe autonomy simulation and distributed training as computationally expensive and say GM has worked to get more simulated miles per GPU-hour (Oct 2026), and its EMWU data-mining pipeline produces embeddings at under $1.00 per million (Jun 2026).

Reported ROI

General Motors has reported no business return, in cost, revenue, headcount or output, that it credits to AI. Mary Barra said on Oct 20, 2026 that every function uses AI and that GM is working to turn that into fixed-cost savings, but declined to quantify them until GM issues 2027 guidance, and Paul Jacobson called GM's AI ambitions in their infancy in Sep 2026.

Productivity markers

General Motors reports process and task-level gains it has not converted into cost: battery packs move through its pack-assembly process 30% faster with PlantBrAIn and digital models (Oct 2026), a sketch-to-animation process that took multiple teams months now takes one designer less than a day, the virtual wind tunnel returns drag in about a minute where a cycle took two weeks (Apr 2026), roof crush analysis runs fell from 8 to 40 hours to under 5 minutes (Jun 2026), and co-simulation work that took months now takes hours or days (Jun 2026). GM research reported that bounded AI refactoring cut code complexity about 30% (Aug 2026), and Paul Jacobson expects AI to flip finance analysts' time from 80% explaining the past to 80% forecasting.

Revenue from AI-enabled services

General Motors sells its in-vehicle AI through OnStar and Super Cruise but does not break out an AI share. OnStar and Super Cruise delivered more than $800 million of recognized revenue in the third quarter of 2026, and GM expects more than $3 billion of realized revenue for 2026 and deferred revenue approaching $7.5 billion by year-end (Oct 2026).

Headcount

General Motors employed about 155,000 people at Dec 31, 2025 (2025 10-K). GM laid off about 600 salaried IT workers, more than 10% of the department, in May 2026 in what TechCrunch described as a skills swap toward AI-native development, data and cloud engineering and agent and model development; GM said it is transforming its IT organization and gave no savings figure. GM also cut about 1,000 software workers in Aug 2024 and CAD execution roles in design engineering in Oct 2025 without tying those cuts to AI, and Mary Barra said in Oct 2026 that AI-driven efficiency would feed significant work on fixed costs.

Governance and risk

Board oversight

General Motors' board oversees enterprise risk and gives its Risk and Cybersecurity Committee responsibility for the cybersecurity program. The 2026 proxy says the board received external presentations on AI focused on cybersecurity, privacy and governance risks, the Audit Committee took AI training in 2025 on how GM and its auditors use the technology, and the Finance Committee reviews disruptive technologies such as AI.

Customer data and consent

General Motors ended its OnStar Smart Driver program in 2024. The FTC proposed an order in Jan 2025 and finalized it in 2026, barring GM from sharing certain driving data with consumer reporting agencies and requiring explicit consent for certain data collection, and in May 2026 GM agreed to pay $12.75 million in a California settlement over the same data sales. Dave Richardson said in Oct 2025 that driver data used by GM's assistant would be opt-in and not sold, and that GM had built a new data team, including Christina Montgomery, IBM's former chief privacy and trust officer, whose GM remit includes data and AI governance. GM says OnStar AI features turn on only with customer consent, with AI permission settings, and that GM cannot access Connected Camera recordings (Oct 2026).

Engineers approve AI output

General Motors' software research limits AI to bounded first drafts that engineers verify, with every interaction scoped so the reviewing engineer can judge it and final approval left with people (Aug 2026). GM's autonomy architecture adds perception outputs and physics and rules-of-the-road guardrails around learned driving models (Mar 2026), and its simulation pipeline pairs AI grading with a human review program called Sim Test/Trust (Oct 2026).

Autonomy safety

General Motors' eyes-off system will use lidar, radar and cameras for perception and start on highways (Oct 2025), and its supervised public-road tests put a trained driver at the wheel who can take control at any time (Mar 2026). GM's 2025 10-K lists risks from accidents, cybersecurity breaches and evolving regulation of autonomous systems, and from access to high-quality data to train the AI models in its driver assistance and autonomy technology.

Disclosed risks and regulation

General Motors' 2025 10-K warns that failing to integrate AI in a timely, cost-effective and compliant way could hurt its strategy, that AI raises the risk of cyberattacks and trade-secret theft, and that the EU Artificial Intelligence Act and other data laws add compliance cost. The 10-K also names the need to attract and retain software talent with AI and machine learning expertise.

Recent signals

Oct 28, 2026
General Motors began a gradual rollout of Google Gemini in Portuguese to Chevrolet Blazer EV, Equinox EV, Equinox Turbo and Silverado vehicles built from 2024 in Brazil, after its Spanish-speaking Latin American markets.
AutoPapo
Oct 21, 2026
At GM Forward 2026, General Motors set a goal of halving vehicle development to about two years with enterprise AI and virtualization, citing more than 300,000 LMR battery designs tested virtually before a physical prototype.
GM News
Oct 21, 2026
General Motors announced the OnStar AI Assistant, due on the 2027 Cadillac Lyriq by over-the-air update in the coming months, and a new OnStar app with AI features from 2027.
GM News
Oct 21, 2026
General Motors extended its planned eyes-off highway driving to full-size SUVs and light-duty pickups and said its autonomy stack drives more than 3 billion virtual miles a day in simulation.
GM News
Oct 20, 2026
Mary Barra said on the Q3 2026 call that every GM function uses AI and that GM would turn that into fixed-cost savings, declining to quantify them before 2027 guidance.
Q3 2026 earnings call transcript
Sep 30, 2026
General Motors described its newly launched OnStar Fleet Intelligence platform, with a built-in Fleet AI Assistant that answers fleet managers' questions in natural language.
GM News
Sep 23, 2026
Paul Jacobson told a JPMorgan conference that General Motors' senior leaders compare monthly AI usage by function and that GM's AI ambitions are in their infancy.
Business Insider
Aug 19, 2026
GM engineers published research on using large language models for test generation, refactoring and code-standard checks in vehicle software, with engineers approving each step.
GM Engineering
Jul 31, 2026
CNBC reported that General Motors is working with an unnamed large language model provider on its own in-vehicle AI assistant, aimed at features such as predictive maintenance and vehicle telemetry.
CNBC
Jun 28, 2026
Rashed Haq told Business Insider that General Motors is hiring autonomy engineers from Tesla, Waymo and Zoox, and a spokesperson said external autonomy hires had nearly doubled from 2025.
Business Insider
Jun 3, 2026
Sterling Anderson and Jason Fischer told Business Insider that General Motors uses AI, simulation and its own engineering data to target a two-year development cycle.
Business Insider
May 11, 2026
General Motors laid off about 600 salaried IT workers, more than 10% of IT, and is hiring for AI-native development, data and cloud engineering and agent and model development, TechCrunch reported, citing a person familiar with the layoffs.
TechCrunch
May 5, 2026
General Motors described machine learning quality tools in its plants, including the WeldBrAIn pilot at Factory ZERO that checks every weld in real time.
GM News
Apr 28, 2026
General Motors announced that Google Gemini would roll out over several months to about 4 million eligible model year 2022 and newer U.S. vehicles with Google built-in.
GM News
Nov 24, 2025
General Motors confirmed that Chief AI Officer Barak Turovsky had left after less than a year and said its AI team would report to the manufacturing engineering organization.
CIO Dive via Yahoo Finance

Quick answers

How does General Motors deploy AI?: General Motors deploys AI mainly in its vehicles and engineering: Google Gemini began rolling out to about 4 million eligible U.S. vehicles in 2026, a GM-built OnStar AI Assistant is due on the 2027 Cadillac Lyriq, and an in-house autonomy stack trains on fleet data and simulation for eyes-off driving in 2028. Inside the company, General Motors tracks AI use monthly by function, applies AI to design, virtual testing and plant quality, and cut about 600 IT jobs in May 2026 while hiring for AI skills.
What does General Motors use AI for?: General Motors' public AI use cases sit in four places: the vehicle and OnStar (Gemini, the OnStar AI Assistant, the fleet assistant, Super Cruise), autonomy development, product engineering (AI design tools, the virtual wind tunnel, virtual crash analysis) and plant quality, and Mary Barra said on Oct 20, 2026 that every GM function uses AI. General Motors has published no business return, such as a cost, revenue or headcount figure, that it credits to AI for any of these use cases; the gains it reports are process speedups, such as faster battery-pack assembly with PlantBrAIn and faster crash and aerodynamic analysis. For AI as a whole General Motors reports no savings figure; Barra said GM could put a number on AI-driven fixed-cost savings when it gives 2027 guidance.
Who leads AI at General Motors?: AI at General Motors is led by Sterling Anderson (Executive Vice President, Global Product, and Chief Product Officer); Rashed Haq (Vice President, Autonomy); Christina Montgomery (Chief Privacy & Trust Officer and Assistant General Counsel). Ownership of AI on the org chart: CPO Sterling Anderson for product AI; AI team under manufacturing engineering since Nov 2025.
What AI platform does General Motors use?: General Motors has named no company-wide employee AI platform; its AI runs as systems it builds in-house, including the OnStar AI Assistant, the PlantBrAIn factory platform, an AI virtual wind tunnel and the Road to Sim autonomy pipeline (Oct 2026).
Which AI models does General Motors use?: General Motors uses Google Gemini in cars with Google built-in, works with an unnamed large language model provider on its own OnStar AI Assistant (Jul 2026), and builds its own vision-language-action and world models for autonomy (Mar 2026).
How many GM employees use AI?: General Motors' AI reaches customers first: about 4 million U.S. vehicles were eligible for Gemini in Apr 2026. Inside GM, Mary Barra's senior leaders compare AI usage by function and department every month, but GM has published no employee user count (Sep 2026).
How far has AI spread at General Motors?: General Motors sits at Piloting: AI runs in production in its vehicles (Gemini for about 4 million eligible U.S. vehicles, Super Cruise), its plants (machine learning quality checks and the PlantBrAIn platform) and its engineering (an AI virtual wind tunnel and AI design tools), and Mary Barra said in Oct 2026 that every function uses AI. General Motors falls short of Scaling on three tests: it has published no company-wide employee AI platform or user count, it has named no successor to Chief AI Officer Barak Turovsky, who left in Nov 2025, and it reports no AI results, with CFO Paul Jacobson calling its AI ambitions in their infancy in Sep 2026.

Open questions

Who leads General Motors' AI team since Barak Turovsky resigned in Nov 2025, what is Behrad Toghi's remit as AI lead, and who in manufacturing engineering does the team report to?
Which AI tools do General Motors employees use, how many people use them, and what do the monthly usage figures by function show?
Which large language model provider works with General Motors on the OnStar AI Assistant, and where does the assistant's model inference run?
Which cloud runs GM's autonomy simulation and training, and is Microsoft Azure still its preferred public cloud after Cruise's robotaxi operations were wound down?
How much fixed-cost saving will General Motors attribute to AI when it issues 2027 guidance?
What business return, in cost, capacity or revenue, do General Motors' deployed AI tools deliver, such as PlantBrAIn, the AI virtual wind tunnel, the design visualization tools, Device Level Analytics, the OnStar virtual assistant and the Fleet AI Assistant?
Who sets General Motors' AI governance rules for employees and vendors, and does Christina Montgomery's privacy and AI governance remit cover internal AI tools as well as vehicle features?
How many vehicles have received Gemini, given that GM put eligible U.S. vehicles at about 4 million in Apr 2026 while a Q2 2026 call transcript quoted about 22 million vehicles receiving a Gemini update?
How many people work on AI and autonomy at General Motors, and how much does it spend on AI compute and model usage?
Which vehicle program will be first to meet the two-year development target, and how much of the reduction will GM credit to AI?

Sources

General Motors

Filings and investor materials

Partners

Microsoft: Cruise and GM team up with Microsoft (Jan 19, 2021)

News

Data confidence

Confirmed: Stated in a primary or reputable public source.
Estimated: Derived or approximated from public reporting.
Unconfirmed: Reported, but not yet corroborated.

General Motors profile · Enterprise AI diffusion report by Imagine · facts checked against their sources on Nov 1, 2026

More company profiles

All profiles