Samsung gives ChatGPT Enterprise and Codex to all employees — what any company can learn

What happened

Samsung Electronics announced it is rolling out ChatGPT Enterprise and Codex to all employees in South Korea and to its entire Device eXperience (DX) division — the part that covers phones, TVs and home appliances — worldwide. Trade press puts the figure at roughly 125,000 employees in Korea alone, which makes this one of OpenAI's largest enterprise deployments so far.

The key point is that this isn't limited access for a pilot team — it's AI made available to the whole organization as a standard work tool. Kim Kyoung-hoon, General Manager of OpenAI Korea, put the difference plainly: Samsung is adopting AI "not as a tool specific for certain teams or tasks but as a core platform to enhance the way employees work and innovate across the organization."

From ban to adoption: what changed in three years

In March 2023, Samsung banned generative AI tools company-wide after a handful of engineers accidentally uploaded proprietary source code and confidential meeting notes to the public version of ChatGPT. It was the natural reaction of a company that lives on intellectual property: if you can't control what goes in and what comes out, you shut it all down.

Three years later, the same company is doing the opposite — not because it stopped caring about data, but because the variant that solves exactly that earlier problem has since arrived: enterprise-grade AI, with contractual data guarantees and clear access rules. The story isn't "Samsung changed its mind," it's "the form in which adoption becomes safe finally showed up."

What employees actually do with ChatGPT and Codex

ChatGPT for knowledge work. Searching and analyzing information, drafting documents, developing ideas, and interpreting data — exactly the tasks that eat time in any office role, regardless of department.

Codex beyond developers. Codex helps developers write, review and debug code, but Samsung is also taking it to non-technical teams — marketing, manufacturing, product designers, corporate functions. In practice: automations and small internal tools built by people who, until now, never wrote a line of code.

It's the same direction we see everywhere in 2026: AI is leaving the IT team and reaching every role that has information to process.

How they handled security

This is the lesson any company can copy, whatever its size. Samsung didn't simply open the tap; it put two layers of protection over adoption:

1. The data guarantee. Enterprise agreements with OpenAI commit, by default, to not training models on customer data. That's the essential difference from the free, public version — the very one that caused the 2023 leak.

2. Mandatory training before access. Access isn't automatic: only employees who complete internal AI security training get it. First you understand what you can and can't put into a model, then you get the tool.

In parallel, the relationship between the two companies is widening: Samsung supplies the advanced memory chips OpenAI needs for its AI infrastructure, and the partnership now extends beyond infrastructure into transforming its own workforce.

Why it matters for your business

The practical takeaway isn't "we too need 125,000 licenses." It's the adoption model: a company obsessed with data protection went from a total ban to AI-for-everyone without taking on the risk it faced three years ago — because it put the rules ahead of the tool. Contractual data guarantee, access granted only after training, control over who works on what.

This is exactly our approach. We build AI agents that do real work — lead qualification, reporting, follow-up, support — but with data hosted in the EU, with logging and human oversight by design, aligned with EU AI Act and GDPR requirements. Adopting AI properly doesn't mean "switch everything on and hope for the best" — it means doing precisely what Samsung did: solve the data and access part first, then scale.

Sources: ↗ Official OpenAI announcement · ↗ The Korea Times