Kakao has established new AI ethics principles and an execution system to apply from service planning to operations as autonomous AI agents spread.

Kakao said on Oct. 15 it unveiled the new AI ethics principles, dubbed SPIRIT, the implementation method NORMA, and related checklists at the developer conference if(kakao)26 held on Oct. 13 at the Kakao AI Campus in Yongin, Gyeonggi Province.

The new ethics principles SPIRIT consist of 6 items: safety, privacy, inclusion, responsibility, user agency, and transparency and trustworthiness.

Kakao used the three core values and seven principles of the Republic of Korea's AI ethics principles as key implementation criteria and supplemented details to fit its service environment.

It named the execution method for applying the ethics principles to actual services NORMA. It is a system that specifies steps to manage and carry out complex ethical tasks based on shared responsibility at the organisational level.

The principles and guide were prepared early this month together with seven external advisers to the AI and Technology Ethics Subcommittee, which was launched in July. The external advisers reviewed consistency with domestic and overseas norms, while inside Kakao, organisations related to AI safety, personal information, information security and services reviewed it from a working-level perspective.

Kakao manages risks that could arise from autonomous agents by dividing them into 5 areas: control leadership, goals, authority, reference resources and interactions.

Each area links necessary control measures such as human oversight and emergency control, authority control and data protection. The AI ethics implementation checklist based on this operates in three steps: identifying service characteristics, pre-screening risks, and checking and determining control items.

Kakao plans to apply the new checklist to actual services within this year.

It is also strengthening checks of AI services for children and teenagers. In addition to the digital child and youth protection checklist it has applied since 2025, Kakao plans to reflect AI-specific risk factors in the new checklist by referring to the OECD and UNICEF child online risk classification system 5Cs.

Kakao has expanded its AI ethics system since announcing an algorithm ethics charter in 2018. In 2022 it established the AI and Technology Ethics Subcommittee with Kakao and major affiliates participating, and later introduced group-wide guidelines for responsible AI, a user agency principle, and a core checklist for safe AI.

Kim Se-woong (김세웅), chair of Kakao's AI and Technology Ethics Subcommittee, said, "By incorporating the perspectives of external advisers and internal organisations, we have established a risk review system that works across the entire life cycle of AI services." He said, "As AI becomes more autonomous, we will clarify human judgment and responsibility and create AI services that users can use with confidence."