Toss Adopts Enterprise ChatGPT License Across Organization

toss

4

  • AI agent–based culture enables team-wide productivity gains, with internal programs in place
  • Advanced security and real-time detection systems allow safer AI usage

Viva Republica, the operator of the financial super app “Toss,” announced that it has officially adopted the “Enterprise Plan License” of ChatGPT, the paid version of the AI service, as part of its initiative to expand the use of artificial intelligence (AI) across the organization.

With this change, all Toss team members can now access the premium ChatGPT service simply by logging in with their work email, without having to request a subscription individually. This move further accelerates Toss' shift toward an AI agent–centered work environment. AI agents are systems based on large language models (LLMs) that can autonomously decide and execute tasks to achieve goals.

Toss has also implemented a support structure to help team members effectively use ChatGPT in their daily work. A dedicated help channel is operated to respond quickly to usage and security-related questions, enabling employees to integrate AI naturally into their workflows.

Additionally, Toss runs an internal program called “LLM Workshop,” where teams can work with in-house LLM experts to apply AI to real tasks. The program helps define possible use cases and design practical implementation strategies, ultimately helping teams develop the capability to turn AI into business impact.

From a security standpoint, Toss has established clear guidelines and technical controls. The Enterprise Plan version of ChatGPT does not use user input for training purposes and includes advanced permission and monitoring tools. On top of this, Toss prohibits entering sensitive information such as source code, infrastructure details, personal customer data, or API keys. If any such input is detected, it is automatically flagged in real time and a follow-up procedure is initiated with the relevant user.

Direct integration between ChatGPT and the internal systems used by Toss or its employees—which operate under a separated network—is strictly prohibited, with limited exceptions under Korea’s Financial Regulatory Sandbox designation. Currently, only 14 specific use cases are allowed for integration, including auto-documentation of code, ad copy review, multilingual translation, and product category prediction.

“This adoption of paid ChatGPT accounts marks a pivotal step in becoming an AI-first organization,” a Toss representative said. “We will continue to build execution strategies that leverage AI agents to drive innovation at speed.”