Naver rolled out new standards that extend its AI model-centered safety management system to the actual process of using services.
Naver Corp. said on the 8th that it unveiled ASF (AI Safety Framework) 2.0, an AI safety system, at the Seoul Forum on AI Safety (SFASS) held in Gangnam-gu, Seoul.
ASF 2.0 upgrades the existing ASF that Naver introduced at the AI Seoul Summit in 2024, focusing not only on checking AI models' performance and risk levels but also on managing safety issues that can arise throughout the entire user journey of using AI services.
Song Dae-seop, Naver AI Safety Policy lead, said in a presentation that "the discussion on AI safety is expanding beyond making individual models safe to the question of how to safely deliver services that combine multiple models to tens of millions of users."
ASF 2.0 reflects changes in the technology and policy environment, including Naver's "on-service AI" strategy, the spread of multi-model environments, and the enactment of the Basic AI Act. Naver classifies risks and conducts impact assessments based on the context of AI services, use cases, and user impact. It also manages the post-launch operations phase through user feedback and continuous safety checks.
Naver also prepared CHEC 2.0, a companywide execution system. The AI Tab released in June went through safety reviews from planning to launch via CHEC 2.0, and the company plans to apply the same system across AI-based services going forward.