South Korea Selects SK Telecom, Kakao and KT to Run Free National AI Service

South Korea has chosen consortia led by SK Telecom, Kakao and KT to build and operate a national artificial intelligence service that will give the public free access to domestically developed AI tools. The selection, which drew six applicants, places three of the country’s largest telecommunications and internet companies at the heart of a government-backed initiative combining local AI models, state-provided computing capacity and consumer-facing distribution.

The Ministry of Science and ICT launched the “AI for All” competition in July, aiming to support a general-purpose chatbot and a public AI agent that can help users find government assistance and complete administrative procedures. Under the program’s terms, selected providers must use South Korean sovereign foundation models for at least half of their services, with at least 30 percent involving domestic models developed by companies other than the operator itself. Foreign models may be used only where minimally necessary, and the government will not subsidize that portion.

To support early development, the government plans to provide the selected companies with 512 Nvidia B200 graphics processing units in 2026. Starting in 2027, it intends to cover costs associated with providing the public service through the national budget, while requiring participating firms to develop revenue models that can sustain service quality over time. The ministry initially targeted a beta rollout around the end of September and a full launch within 2026. SK Telecom’s consortium now says its own beta is planned for October, followed by a formal launch before year-end; exact rollout dates for the Kakao- and KT-led services were not disclosed.

In its operator announcement, SK Telecom said its service will use its A.X K2 foundation model alongside other South Korean AI models. The assistant will be available through an app as well as phone calls and text messages, an approach aimed at older people and users less comfortable with app-based services. The system is designed to go beyond answering questions by helping users complete tasks, including requesting 83 types of official certificates through Government24 and PASS, explaining health records, suggesting nearby medical services, tracking job openings and deadlines, and guiding small businesses through tax information and government support programs.

Because the assistant could initiate calls, reservations and other actions, SK Telecom said its design will include user confirmation and approval, alongside measures to minimize the personal data used. These features remain development plans rather than demonstrated capabilities, and their final scope will depend on technical integration, privacy controls and government approval. The company said 20 organizations are participating directly in its consortium across financial services, mobility, healthcare, education and care services, with memoranda of understanding signed with 11 additional organizations, bringing the broader group to 31 companies and institutions.

The selection gives South Korea three competing channels for a population-scale AI rollout. SK Telecom can draw on its mobile network and voice technology, Kakao can reach users through its widely used messaging ecosystem, and KT can distribute services across its telecommunications and digital platforms. However, the government has not publicly disclosed evaluation scores or rankings of the six applicants, and detailed official plans from the Kakao and KT consortia were not available in primary materials reviewed for this article. Their model mix, interface, data-handling rules and launch schedule will be important points to confirm as the program moves toward beta testing.

The initiative represents a major test of whether government-supported domestic AI can be delivered as a broadly accessible public utility. It also raises practical questions around user eligibility, data governance, service limits and the reliability of AI systems used to navigate health, benefits and administrative processes. Those issues are likely to shape public adoption as each operator moves from proposal to live service.