Three years ago, Samsung Electronics banned generative AI from its offices. This week, it began putting the technology back into employees’ hands — this time with enterprise guardrails.
In one of the largest enterprise deployments OpenAI has ever announced, Samsung is rolling out ChatGPT Enterprise and Codex to all of its employees in South Korea and to every worker in its global Device eXperience (DX) division, the sprawling business behind Galaxy phones, televisions, and home appliances. Announced in late June, the move puts premium AI tools in the hands of a vast cross-section of Samsung’s workforce across research labs, manufacturing operations, marketing teams, and back offices.
It also marks a striking about-face for a company that, in 2023, became a cautionary tale about the risks of workplace AI.
Samsung's reversal is all the more remarkable given how it began. In the spring of 2023, engineers accidentally fed sensitive internal source code and meeting notes into ChatGPT. The company's response was swift and absolute: a blanket ban on generative AI tools.
What changed wasn't the risk. It was the guardrails. Rather than turning employees loose on an open consumer chatbot, Samsung has been building the policies, training, and security systems needed to use AI more safely. ChatGPT Enterprise, OpenAI's business tier, keeps company data walled off from model training by default and layers on the access controls and governance framework a corporation of Samsung's size demands. Over a two-month pilot in spring 2026, the DX division put roughly 2,500 employees through trials of ChatGPT, Google's Gemini, and Anthropic's Claude. Samsung is also pairing the rollout with internal AI training, with broader employee training scheduled through the end of 2026.
The pitch to employees is breadth. Samsung plans to apply ChatGPT and Codex to both technical and non-technical work: software development, product design, marketing, manufacturing, and the everyday grind of corporate life. For knowledge workers, that means leaning on ChatGPT to sift through information, draft documents, develop ideas, and make sense of data.
Codex, OpenAI's coding agent, is the other half of the rollout. It started life as a tool for writing code, then reviewing and debugging it. Increasingly, though, it's used beyond traditional software teams: helping employees spin up internal tools and simple websites, or automate repetitive workflows.
The appetite is already visible.
OpenAI says more than five million people now use Codex every week, and in South Korea, weekly active users have jumped roughly 800% since February. That surge preceded the full Samsung rollout and hints at how fast adoption can compound inside a tech-forward economy.
Executives on both sides framed the deal as something bigger than a software license. Roh Tae-moon, who heads Samsung's DX division, cast it as a structural shift, saying, "This is not simply about introducing AI as a workplace tool." OpenAI's Korea chief, Harrison Kim, made a similar point, characterizing the deployment as Samsung making AI a foundational part of how its people work, rather than a perk reserved for a few teams.
That framing reflects a broader bet. The rollout is part of an internal "AX," or AI Transformation, initiative. Notably, OpenAI isn't the only vendor in the mix. Samsung is running a multi-vendor stack, giving employees access to Gemini and Claude alongside ChatGPT so teams can pick the right model for the task at hand.
There's a hardware subplot, too. Samsung and OpenAI were already partners before this deal: Samsung supplies advanced memory chips used in the data centers behind AI models, and the companies announced a 2025 agreement tied to OpenAI's Stargate infrastructure project. With this week's news, that partnership now runs in both directions — Samsung helps supply the silicon needed to build AI infrastructure, and buys OpenAI's software to reinvent how it works.
For years, enterprise AI lived in pilot programs and proofs of concept: a handful of teams running controlled experiments while everyone else waited. Samsung's deployment signals a different phase. When one of the world's largest manufacturers moves a flagship AI product from a test group to a major part of its workforce, and reverses a high-profile ban to do it, the message to other large companies is hard to miss.
The open questions are real. Samsung's semiconductor division still faces tighter restrictions, and the financial terms of the deal haven't been disclosed. Whether the security training and access controls hold up at full scale is the test that matters now. But the direction is clear: enterprise AI is moving from cautious pilots into everyday corporate work, and Samsung is betting that the productivity payoff will be worth it.
Comments