In South Korea, the telecom industry is also stepping up efforts to prepare countermeasures in line with the AI Framework Act, the world's first of its kind to take effect. Companies are accelerating companywide governance overhauls, including organisational changes, beyond revising service terms and interfaces.

The AI Framework Act, which took effect on Jan. 22, stipulates transparency obligations for high-impact AI and generative AI. That increases the burden on companies that provide AI services. Telecom operators pushing a shift into AI companies are focusing on building systems for legal compliance and risk management.

◆ Stepping up companywide governance overhaul… applying ethics principles

SK Telecom strengthened its companywide AI governance framework to match the AI Framework Act. From Jan. 22, when the law took effect, it started an internal campaign called "Good AI" and shared with employees a summary of key provisions of the act and privacy compliance requirements. It also upgraded processes for using its "AI Governance Portal", which it opened in September last year.

The AI Governance Portal is particularly useful in analysing risk and opportunity factors in AI services. For example, when a developer starts a new AI project, the developer registers it on the portal and the system automatically asks questions. It asks whether the AI handles customer data, whether it is used in healthcare or finance, and whether it could generate deepfakes. Based on the answers, it assigns a risk rating and creates a checklist accordingly. The higher the risk, the more items there are to review.

In December last year, it also set up a system in which the chief privacy officer (CPO) oversees AI governance. Cha Ho-beom (차호범), SKT's CPO, said, "We will bring together companywide capabilities to develop and use AI that is responsible and trustworthy." He added, "We will provide better value to customers through safe AI."

KT established its own AI ethics principles called ASTRI and is applying five core principles across the entire process from AI planning, development, operation and use: responsibility, sustainability, transparency, reliability and inclusivity. On the technical side, it reviewed domestic and overseas AI risk management frameworks and established its own AI safety standards.

In September last year, it also released "AI Guardrail (SafetyGuard)", which blocks harmful responses from AI models in real time, on the Hugging Face platform. The company's goal is to support customers so they can pursue AI transformation with confidence through trustworthy AI. A KT official explained that it also published a report on responsible AI technologies, presenting a lifecycle risk management framework from development to operation, an evaluation framework applicable to real-world environments, and implementation strategies.

LG Uplus operates an AI risk management framework involving relevant departments such as the chief technology officer, the information security centre and the legal office. It manages the entire process of AI service planning, development and operation in a systematic way to ensure legal compliance. It also strengthened training to help employees understand and comply with relevant laws. An LG Uplus official stressed that, with the AI Framework Act now in force, it is a time when responsible use based on customer trust is required.

◆ "Beyond regulatory response, it will become a 'trust competition'"

In the industry, there is a view that the enforcement of the AI Framework Act will become a turning point for a trust competition among telecom operators, going beyond a simple regulatory response. The analysis is that efforts are needed to block more risks and variables as companies accelerate their shift to AI businesses.

Telecom operators already run a range of AI services including AI consulting, recommendations and content generation. The more customer touchpoints expand, the larger the risks become, including malfunctions, bias and privacy violations. Telecom operators also serve tens of millions of subscribers, meaning they are required to meet both public interest and social responsibility. In this structure, advance checks and internal control systems for AI services are directly linked to corporate credibility.

In particular, after suffering a hacking incident last year, any controversy related to AI would inevitably damage their image. As a result, voices in the industry say it is essential to establish norms and systems for practices that go beyond legal standards.

One industry official said, "Legal compliance is only a basic condition." The official added, "As the scope of AI use expands, risk management and securing transparency will become a company's brand image."