Samsung Deploys ChatGPT Enterprise and Codex Companywide, Years After Its AI Ban

Years ago, Samsung became the cautionary tale. A proprietary source code leak tied to employee use of generative AI tools led the company to ban ChatGPT and similar tools internally, a decision that rippled across enterprise AI adoption conversations worldwide. Today, OpenAI’s announcement frames Samsung as the comeback story: a full enterprise deployment to all Korean employees and its entire Device eXperience (DX) division globally, with a governance model designed to prevent the problem that triggered the ban. The deployment covers all Samsung Electronics employees in South Korea and all employees in Samsung’s Device eXperience division globally, according to OpenAI’s announcement. Samsung SDS is listed among the companies in Korea also using ChatGPT Enterprise, OpenAI APIs, and Codex, per the same announcement. ChatGPT Enterprise provides enterprise-grade capabilities including data protection, user and access management, and security controls, which OpenAI says allow Samsung employees to use advanced AI within the company’s security policies and governance framework. Whether that governance model is sufficient in practice is unverified — it’s an announced design, not an independently audited one. OpenAI also reported that weekly active Codex usage among Samsung employees in South Korea grew by nearly 800% since February 1, 2026. Both figures are drawn from OpenAI’s own deployment data and haven’t been independently confirmed. Why this matters for enterprise AI practitioners. Samsung’s earlier ban was cited repeatedly by enterprise IT and legal teams as justification for blocking or delaying generative AI rollouts. The reversal, and specifically the security-and-governance-gated access model, gives those teams a concrete framework to point to. The structure is replicable: defined scope by division and geography, enterprise-grade security controls, and broad functional coverage from R&D and manufacturing to marketing and corporate functions. Whether the controls actually prevent data leakage at scale is the open question. That answer won’t come from the announcement. The competitive dynamics are also worth noting. This isn’t just a single enterprise deployment; it’s infrastructure for regional distribution. What to watch. The 800% Codex growth figure points to developer-side adoption outpacing general enterprise use. If Samsung’s internal developer community drives disproportionate usage, expect that pattern to show up in how other large enterprises structure their AI rollouts — Codex-first, developer-gated, before broader employee access. Watch for Samsung’s disclosure of any data governance incidents, which would be the first independent signal on whether the governance model works. Watch also for competitive responses: if Samsung’s model becomes a reference architecture, expect Microsoft Copilot and Google Workspace AI to position their own enterprise governance frameworks against it. TJS synthesis. The governance model is the story, not the scale. The companywide scale is notable. But the reason this deployment matters is that Samsung solved, or at least addressed, the trust problem that made enterprise AI adoption a legal and compliance risk. Enterprise-grade security controls, defined organizational scope, and broad functional rollout are defensible design choices. Enterprise teams that have been waiting for a credible reference architecture now have one. Don’t treat the 800% growth figure as evidence it’s working — it’s a vendor-reported adoption metric. The real test is what happens at month six of full deployment, when the governance pressure is real.