Samsung is turning from restriction to controlled deployment. OpenAI says Samsung Electronics is making ChatGPT Enterprise and Codex available to all Samsung Electronics employees in Korea and to all employees worldwide in its Device eXperience division.
The Economic Times reported the move on June 23, 2026 as a reversal from Samsung’s 2023 curbs on employee ChatGPT use after confidential-information concerns. OpenAI’s own announcement says the rollout is one of its largest enterprise deployments to date.
What changed
- Samsung is deploying ChatGPT Enterprise and Codex to employees around the world.
- Coverage includes all Samsung Electronics employees in Korea and global DX employees.
- OpenAI says Samsung plans to use the tools across software development, marketing, product development, manufacturing, corporate work, and other functions.
- OpenAI says Codex has more than 5 million weekly users and that weekly active Codex users in Korea have grown nearly 800 percent since February 1, 2026.
- The deployment expands an existing OpenAI-Samsung relationship beyond AI infrastructure and memory semiconductors into workforce adoption.
Buyer value
This matters because the enterprise AI debate is moving past “ban it” versus “let everyone use it.” Large companies are now looking for controlled internal deployments with admin policy, access management, data controls, and narrower tool scopes.
For buyers, Samsung is a useful benchmark, not a template to copy blindly. The lesson is to build a rollout plan around data classes, teams, workflows, and model access. A coding agent needs different controls from a writing assistant.
Ask before signing:
- Which employees get ChatGPT Enterprise, Codex, or both?
- Are source code, customer data, design files, and manufacturing data governed separately?
- Which admin logs are available to security and compliance?
- Can the company disable features by team, geography, project, or data class?
- Does the contract preserve data isolation, retention rules, and audit rights?
- What happens when a model, route, plugin, or app feature changes?
What to do
If you are replacing an internal AI ban, do not start with a company-wide memo. Start with three workflows: knowledge work, coding, and internal automation. Define approved data, blocked data, monitoring, and exception paths for each one.
If you already use ChatGPT Enterprise, this is a reminder to inspect Codex separately. Developer agents can read repositories, propose code, run commands, and create artifacts. That is a different risk surface from chat.
If you are comparing ChatGPT Enterprise with Claude, Gemini, Copilot, or an internal model gateway, ask for proof of enterprise controls rather than vendor slogans. The rollout plan is the product.
AiPedia take
Samsung’s move is a major adoption signal. The valuable part is not that a famous company is using ChatGPT. The valuable part is that enterprise AI governance is shifting from blanket restrictions to controlled access, workflow-specific policy, and coding-agent oversight.
Sources
Primary and corroborating references used for this news item.