Three years after banning ChatGPT following a high-profile data leak, Samsung Electronics has reversed course in a way that signals something important about where enterprise AI adoption now stands. The company announced a full global deployment of ChatGPT Enterprise and Codex to its employees, making it one of the largest enterprise rollouts in OpenAI’s history.

The rollout covers all employees in Samsung’s Korean operations and all employees globally within its Device eXperience (DX) division.

What Happened in 2023

In 2023, a Samsung engineer pasted proprietary chip design data into ChatGPT, triggering an internal review that concluded within weeks with a company-wide ban on external AI tools. The story became one of the most cited examples of AI risk in enterprise settings. Consultants built slides around it. Security teams used it to justify blanket prohibitions.

The logic was understandable at the time. Data governance frameworks for AI tools did not yet exist in most large organisations. The tooling had no real enterprise controls. Blocking was the only reliable option.

That calculus has changed.

Why Samsung Changed Its Mind

ChatGPT Enterprise now ships with features that did not exist when Samsung issued its ban: data protection controls, user and access management, security audit logs, and admin dashboards that let IT teams see usage patterns without inspecting content.

Samsung ran a two-month internal pilot with around 2,500 employees, testing AI tools inside a secure sandbox before committing to the wider rollout. Employees who completed mandatory security training received access. Those who did not, did not.

The result: a phased, governed deployment rather than an open-ended experiment.

The Codex Number Worth Watching

The headline metric in the announcement was not about ChatGPT. It was about Codex.

Codex weekly active users in Korea grew nearly 800% since February 1, 2026. That is a significant number because Codex is not purely a developer tool. Samsung is explicitly deploying it to non-technical teams to help with day-to-day work, not just writing and debugging code.

That matters because it suggests something the broader market is still catching up to: AI coding agents are beginning to escape the engineering function. Business analysts, operations teams, and product managers are using tools like Codex to build lightweight automations, query data, and create internal apps without raising an IT ticket.

Over five million people now use Codex weekly globally across all enterprise deployments.

What This Means for Business

Samsung’s decision tells a story about the maturation of enterprise AI governance. The fact that a company famously burned by an AI data leak three years ago is now one of OpenAI’s flagship enterprise customers shows how far the controls have come.

For business leaders still sitting on the sidelines citing risk concerns, the Samsung example removes one of the most common objections. Enterprise-grade AI tools now come with audit trails, access controls, and security policies that align with how large organisations actually operate.

But the Codex adoption curve reveals something equally important: the ROI case for AI tools is moving beyond the developer function. When non-technical employees are adopting a coding agent at this pace, it suggests the productivity gains are real and visible enough for individuals to seek the tool out themselves rather than waiting to be directed.

The companies that capture the most value from AI in the next 18 months will not be the ones that deployed the fastest in 2024. They will be the ones that paired deployment with the training and governance infrastructure to make the tools actually useful to the people using them.

That is the part most enterprise AI announcements leave out. Samsung’s phased pilot and mandatory training model is a better blueprint than the typical “we gave everyone access” rollout.

At Enterprise DNA, we spend a lot of time on this gap: the difference between organisations that buy AI access and organisations that build AI capability. The Samsung story illustrates exactly why that distinction matters.