KT Cloud, Kakao team up to build safer generative AI environment

KT Corp's cloud unit kt cloud announced Friday that it has established an AI safety cooperation framework with Kakao to enable secure AI service deployment, with both companies committing to build a trustworthy AI environment across the public and private sectors.

As generative AI adoption expands, concerns over AI safety — including harmful content generation, prompt injection attacks and personal data leaks — have grown significantly. Public institutions, financial firms and corporate customers must consider not only performance but also security, reliability and regulatory compliance when adopting AI, driving demand for safer AI operating environments.

kt cloud and Kakao agreed to combine kt cloud's AI infrastructure platform capabilities with Kakao's AI safety technology, laying the groundwork for customers to develop and operate generative AI services with confidence. The two companies plan to contribute to the safe expansion of the domestic AI ecosystem through technical cooperation aimed at enhancing the safety and reliability of AI services.

The agreement was signed Wednesday afternoon at Kakao's Pangyo office, where Kong Yong-jun, head of kt cloud's cloud division, and Kim Se-ung, Kakao's AI synergy performance leader, discussed the details of the AI safety partnership and its implementation roadmap.

As a first step, kt cloud will integrate Kakao's AI guardrail model, Kanana Safeguard, into RAG Suite 2.0 — a retrieval-augmented generation-based generative AI service platform for public-sector customers that kt cloud launched in April.

Kanana Safeguard is an AI safety model tailored to the Korean language and Korea's cultural and legal context. It detects and blocks harmful content generation, inappropriate responses and prompt injection attacks to support the safe use of generative AI services.

Beyond the guardrail integration, the two companies plan to roll out a range of AI safety tools to kt cloud customers in phases, including a model stability evaluation system and an AI red-teaming system that diagnoses vulnerabilities based on various attack scenarios.

The partnership will also extend to an operations platform that integrates and manages AI safety tools using customer feedback and operational data collected within the kt cloud environment. Customers will be able to build a more systematic AI operating environment covering the full lifecycle — from development and deployment to monitoring — while more effectively meeting the security and compliance requirements of public institutions and financial sector clients.

kt cloud said it will leverage the technical cooperation with Kakao to help public and private customers build a more trustworthy AI environment.

"As generative AI moves into full-scale adoption, establishing a foundation for responsible AI use — not just high performance — has become a critical challenge," said Kim Bong-gyun, chief executive of kt cloud. "Through our partnership with Kakao, we will combine AI infrastructure with safety technology to support public and private customers in developing and operating AI services in a more reliable environment."