KT Strengthens ‘Responsible AI’ Framework, Accelerates AX Support
As artificial intelligence (AI) technology becomes more sophisticated and demands for ethics and reliability grow, KT is accelerating its support for AI transformation (AX) in corporate and public sectors by promoting its responsible AI framework.
On January 21, KT announced that it is continuously strengthening its responsible AI framework and dedicating efforts to building an AI environment that companies and society can utilize with confidence.
The Evolution of AI
Recently, AI has evolved beyond simple content generation tools into agentic AI that understands objectives, makes independent judgments, and performs tasks. As AI’s involvement in decision-making and execution processes expands, industries are experiencing growing expectations for technological innovation alongside increasing demands for ensuring AI safety and reliability.
Establishment of Ethical Frameworks
In response to these environmental changes, KT has established AI ethics not merely as a compliance dimension but as a core value for leaping toward AICT (AI+ICT)-centered management, consolidating company-wide frameworks.
KT established the dedicated organization Responsible AI Center (RAIC) in 2024 and became the first domestic telecommunications company to appoint a chief responsible officer (CRAIO). Additionally, it established its own AI ethics principles called ASTRI, applying five core principles throughout the entire AI planning, development, operation, and utilization process: accountability, sustainability, transparency, reliability, and inclusiveness.
Reporting and Transparency
KT also publishes its performance and AI ethics strengthening activities through the annually released KT Responsible AI Report. The report published in December last year detailed KT’s self-developed AI model’s social impact and safety evaluation processes, and AI guardrails (SafetyGuard). It introduced practical cases such as RAI education expansion.
External Collaboration and Education
KT continues to operate the Responsible AI Advisory Committee composed of external experts and participates in AI-related institutional discussions through cooperation with the government and industry. It is also working to spread a responsible AI culture by operating AI ethics education as mandatory training for all employees, including subsidiaries and partner companies.
Technical Standards and Innovations
On the technical side, KT reviewed domestic and international AI risk management frameworks to establish its own safety standards suitable for the domestic environment. In October last year, it published the Responsible AI Technical Report containing a full-cycle management framework for AI risk identification, evaluation, and mitigation, presenting evaluation frameworks and implementation strategies applicable to actual service environments.
The “AI Guardrail” that blocks harmful responses from AI models in real-time was released on the global AI platform Hugging Face in September last year, achieving an F1 score of 97 points in Kor Ethical QA, a Korean AI ethics response quality evaluation standard.
Recognition and Certifications
As a result of these efforts, KT’s self-developed AI model Mideum K 2.0 Base became the first among domestic AI models to obtain AI Reliability Certification 2.0 (CAT 2.0) from the Telecommunications Technology Association (TTA). This certification is significant as it represents the first case of applying an advanced certification framework based on international standards, objectively verifying accountability and reliability throughout AI model development and operation.
KT’s responsible AI framework is also being recognized on the global stage. The GSM Association (GSMA) evaluates RAI maturity in four stages, and KT achieved the highest grade “Advanced” (corporate culture internalization) in the 2025 evaluation, becoming the first domestic mobile telecommunications company to be introduced as a global best practice.
Future Plans
KT plans to support domestic companies and public customers in pursuing AI transformation with confidence through trustworthy AI and readily applicable AI, beyond performance-centered AI.
KT CRAIO Executive Managing Director Bae Soon-min stated, “We will continue to contribute to the healthy growth of the domestic AI industry and securing user trust based on reliable AI technology.”
