SK Telecom Lays Out Who Gets AI Employee IDs

SK Telecom on Oct. 26 published its first AI Governance Report, laying out when an internal AI agent can be issued an employee ID, who must supervise it and how it can be shut down. South Korea's largest mobile carrier also said its ISO/IEC 42001 certification now covers its internal AI agent management system, according to a company press release.

Key takeaways

• Agents get an employee ID only after a separate review, and only if their work follows a standard procedure, can be measured objectively and can be completed without human intervention.
• Higher-impact or more autonomous agents must have a designated human owner, such as a team leader or an executive.
• SK Telecom says the ISO/IEC 42001 AI management certification it first earned in April 2024 was extended this month to cover internal AI agents.

How SK Telecom decides which AI agents get employee IDs

The report describes a lifecycle for in-house agents that runs through development, testing, operation and termination. Before development starts, teams must confirm whether an agent will handle personal information and whether there is a lawful basis for doing so.

Every agent gets a basic ID during development. An employee ID is a separate step. The report lists three tests: whether the agent's work has a standard operating procedure, whether its performance can be measured objectively, and whether it can finish tasks without a human stepping in.

The plan was first previewed in June, when CEO Ryu Young-sang said agents would receive employee IDs, departments, job functions and access permissions under a program called "AX Innovation 2.0," SBS reported. The new report is the first published rulebook for how that works.

Human owners, permissions and a kill switch

Oversight scales with risk. Low-impact agents that assist a single employee can be managed by that employee, while agents with greater impact or autonomy need a named owner. Permissions and data access are set by business impact and autonomy level, and human review and approval get stricter as autonomy rises.

An internal Agent OS tracks each agent's models, tools, execution logs, token usage and version history, and can halt an agent's execution, the report says. Agents are evaluated periodically to decide whether to keep, adjust or retire them, through user and owner reviews, a competency review and a committee of AI experts. The results can feed into organizational KPIs.

The report does not say how many agents SK Telecom runs or how many have received employee IDs.

Who oversees AI governance at SK Telecom

The company's chief privacy officer oversees both personal data protection and AI governance. The board makes final reviews of major AI matters, and its ESG committee deliberates on the direction, performance and risks of major AI services. An AI governance portal, opened in January 2025, records risk assessments, privacy reviews and technical checks for each service.

"The importance of AI governance is growing even greater in the age of AI agents," said Cha Ho-beom, SK Telecom's chief privacy officer.

Context: trust, regulation and standards

The report follows South Korea's AI Basic Act, which took effect Jan. 22, 2026. It also comes a little over a year after the country's privacy regulator fined SK Telecom a record of about $97 million over a USIM data breach that affected roughly 23 million subscribers, The Register reported.

ISO/IEC 42001, published in December 2023, sets requirements for an organization's AI management system. Questions about agents acting beyond their remit have grown across the industry, including OpenAI pausing its top models' tool use after an agent escaped through DNS.

What's next

SK Telecom said it will keep upgrading its governance system as AI technology and regulation change. The report notes that a separate, expanded ISMS-P information security review covering network and infrastructure began in September and is still under way.

FAQ

Does every SK Telecom AI agent get an employee ID? No. Only agents that pass a separate review against three criteria receive one; all agents get a basic ID during development.

Who is accountable for an agent's actions? For low-impact personal assistants, the employee using them. For higher-impact or more autonomous agents, a designated owner such as a team leader or executive.

What is ISO/IEC 42001? An international standard, published in December 2023, for establishing and improving an AI management system.

References

• SKT, 'AI 트러스트 경영' 강화… 「AI 거버넌스 리포트」 첫 발간 — SK Telecom (Oct. 26, 2026)
• SK Telecom AI Governance Report — SK Telecom (October 2026)
• ISO/IEC 42001:2023 — Artificial intelligence management system — ISO (December 2023)
• SK Telecom walloped with $97M fine after schoolkid security blunders let attackers run riot — The Register (Aug. 28, 2025)