KT (030200.KS) has proposed a new international standard for developing cybersecurity-focused artificial intelligence using general-purpose AI models.
The company said on the 21st that it attended an international standards meeting of Study Group 17 (SG17), the security study group under the International Telecommunication Union's Telecommunication Standardization Sector (ITU-T), held in Chongqing, China, from Sept. 7 to 11, where it proposed standardization measures for AI cybersecurity and digital trust. ITU-T SG17, overseen by the ITU, a United Nations agency, develops international standards in information security, including cybersecurity, personal data protection and AI security, with participation from government officials, companies and research institutes worldwide.
At the meeting, KT proposed guidelines for developing cybersecurity-specialized AI models based on general-purpose foundation models as a new international standard work item. The proposal covers cybersecurity knowledge and data for training AI models, technical factors to consider when developing security-focused models, and evaluation criteria across the entire development process.
Once the standard is established, companies and institutions will be able to develop and operate cybersecurity AI models under common criteria. It is also expected to improve interoperability among AI-based threat detection, analysis and response technologies.
KT also took part in international discussions on digital trust in the AI era. The company proposed revisions to refine key terminology used in SG17's technical report on trust frameworks and to add components and requirements for such frameworks. The revisions were adopted with support from the United Kingdom, India and other countries.
Kim Chang-oh, senior vice president and chief privacy officer at KT's information security division, chairs SG17's working party on trust and security strategy and coordination (WP3) and is leading the standardization discussions. Yeom Heung-youl, chairman of KT's privacy advisory committee, who served as SG17 chairman from 2016 to 2024, also joined the standards development work. "As AI technology spreads across industries, it is important for the international community to jointly establish development criteria for AI models specialized in cybersecurity and trust frameworks for ICT services," Yeom said.
