Samsung Electronics has deployed ChatGPT Enterprise and Codex across its entire Korean workforce and all employees in its Device eXperience (DX) division worldwide. This is one of OpenAI's largest enterprise AI deployment deals to date, and it directly reverses Samsung's company-wide ban on generative AI tools enacted in March 2023 after employees leaked confidential source code through ChatGPT.
The reversal signals something concrete: the enterprise security concerns that kept large manufacturers away from generative AI tools are now addressable through proper access controls, data isolation, and administrative governance layers.
TL;DR
- Samsung has rolled out ChatGPT Enterprise and Codex to all employees in Korea and its global DX division, one of OpenAI's largest enterprise deals.
- This reverses Samsung's 2023 ban on generative AI, imposed after engineers leaked sensitive code via ChatGPT.
- Codex is being used for both technical work (code review, debugging) and non-technical tasks (building internal tools, automating workflows).
- OpenAI reports 5 million+ weekly Codex users globally, with Korean usage growing roughly 800% since February 2026.
- Samsung is running a multi-vendor AI stack alongside other providers, not betting on a single platform.
Why Did Samsung Ban ChatGPT in the First Place?
In March 2023, Samsung engineers pasted confidential source code and internal meeting notes directly into ChatGPT. The data left Samsung's control entirely. The company responded with an immediate, company-wide prohibition on generative AI tools.
This incident became shorthand across the industry for "why enterprises cannot trust public AI." It was cited in countless boardroom discussions as justification for blanket bans.
The path back took roughly three years. In October 2025, Samsung and OpenAI signed a strategic partnership covering AI infrastructure, including memory semiconductors and data centre collaboration. By late 2025, Samsung SDS became the first Korean entity authorised to manage ChatGPT Enterprise deployments. The full workforce rollout followed in June 2026.
What Does the Enterprise AI Deployment Actually Include?
Samsung's deployment covers two products:
ChatGPT Enterprise provides the conversational AI layer with enterprise-grade data protection. User and access management controls let administrators enforce policies about what data enters the system. Crucially, enterprise-tier agreements guarantee that customer data is not used for model training — the exact gap that caused the 2023 incident.
Codex handles the autonomous software development layer. For Samsung's engineering teams, this means code writing, review, and debugging assistance. For non-technical staff in marketing, product development, and manufacturing, Codex enables building internal tools, websites, and automated workflows without writing code manually. Our detailed guide to Codex as an autonomous worker covers the technical architecture behind this.
Samsung plans to use these tools across software development, marketing, product development, and manufacturing — not limiting AI to IT departments alone.
How Did Samsung Solve the Security Problem?
The gap between 2023's ban and 2026's deployment was not simply waiting for better technology. It required structural changes:
- Data isolation guarantees — ChatGPT Enterprise contractually separates customer data from training pipelines.
- Administrative controls — Samsung can enforce which employees access which capabilities, audit usage, and revoke access centrally.
- Authorised reseller model — Samsung SDS manages the deployment internally, keeping procurement and governance within Samsung's own infrastructure group.
- Multi-vendor diversification — Samsung reportedly deploys multiple AI platforms (including solutions from Google and Anthropic), reducing single-vendor dependency.
For organisations still operating under blanket AI bans, Samsung's trajectory provides a template: partner with a trusted internal entity for governance, negotiate enterprise-grade data agreements, and deploy incrementally rather than all-at-once.
If you are evaluating platforms for your own organisation, our comparison of AI automation platforms for business covers the current options.
What Does 800% Growth in Korean Codex Usage Mean?
OpenAI reports that weekly active Codex users in Korea grew approximately 800% between February and June 2026. Globally, more than 5 million people now use Codex weekly for both technical and non-technical workflows.
These numbers matter because they indicate that AI coding assistants have moved from developer experimentation into daily enterprise infrastructure. The growth is not limited to Samsung; other Korean firms using ChatGPT Enterprise include LG Electronics, LG Uplus, LG CNS, GS E&C, Krafton, Toss, and Korea Zinc, among others. Seoul National University has also deployed ChatGPT Edu to all 47,000 community members.
This regional clustering suggests that once enterprise AI deployment reaches critical mass in a market, adoption accelerates rapidly as competitive pressure compounds. For more on how Codex fits into automating business tasks, we have covered the practical applications in detail.
What Should Other Enterprises Learn from This?
Samsung's reversal is instructive, but it is not a universal green light. Several conditions enabled this move:
- Samsung has an internal technology services arm (Samsung SDS) capable of managing deployment governance.
- The company had three years of internal policy development between ban and rollout.
- Enterprise-tier AI products now offer contractual data protections that did not exist in early 2023.
- Training is expected to continue through end of 2026, meaning Samsung is treating this as a gradual capability build, not a switch-flip.
Organisations without Samsung's internal IT governance infrastructure will need external partners or managed service layers to achieve equivalent control. The OpenAI agent stack guide explains how the underlying APIs and SDKs connect for custom enterprise integrations.
It is also worth noting Samsung's multi-vendor approach. Betting entirely on one AI provider carries concentration risk — both technically and commercially. Understanding OpenAI's business model and financial trajectory helps when evaluating long-term vendor stability.
FAQ
Q: When did Samsung deploy ChatGPT Enterprise and Codex? A: OpenAI announced the deployment on 21 June 2026. ChatGPT Enterprise and Codex are available to all Samsung Electronics employees in Korea and all employees in the global Device eXperience (DX) division.
Q: Why did Samsung originally ban ChatGPT? A: In March 2023, Samsung engineers inadvertently pasted confidential source code and internal meeting notes into ChatGPT, causing a data leak. Samsung imposed a company-wide ban on generative AI tools in response.
Q: Does Samsung use only OpenAI products for AI? A: No. Samsung reportedly operates a multi-vendor AI stack that includes solutions from multiple providers. The ChatGPT Enterprise deployment is one component of a broader AI strategy.
Q: Is Samsung's employee data used to train OpenAI's models? A: No. ChatGPT Enterprise agreements guarantee that customer data is not used for model training, which was a core requirement for Samsung's deployment.
Q: How many people use Codex globally? A: OpenAI reports more than 5 million weekly active Codex users as of June 2026, spanning both technical and non-technical workflows.
Discussion
0 comments