NAVER Unveils AI Safety Framework 2.0, Expanding Safety Management Scope to Cover Users and Services
NAVER Unveils AI Safety Framework 2.0, Expanding Safety Management Scope to Cover Users and Services
NAVER Unveils AI Safety Framework 2.0, Expanding Safety Management Scope to Cover Users and Services
- ASF 2.0 unveiled at the Seoul Forum on AI Safety & Security on July 8, reflecting NAVER’s On-service AI strategy and the latest global trends
- Safety management scope expanded to focus on users and services, with more detailed classification and impact assessment criteria
- ASF 2.0 applied to AI Tab, with safety reviews conducted throughout the service lifecycle and expanded external collaboration
July 8, 2026
NAVER Corp. (CEO Choi Soo-yeon) unveiled NAVER ASF (AI Safety Framework) 2.0 on July 8, expanding its AI safety management scope to help ensure safe AI service experiences for users.
ASF 2.0 is NAVER’s updated AI safety management framework, building on the ASF unveiled at the 2024 AI Seoul Summit and extending it to encompass user protection. Whereas the original ASF primarily addressed the performance and risk levels of AI models, ASF 2.0 places the services users actually experience at the center of safety management.
At the Seoul Forum on AI Safety & Security (SFASS), held in Gangnam, Seoul, on July 8, NAVER outlined the direction and key components of ASF 2.0, including its enhanced classification, management and implementation systems, and shared its plans for future initiatives.
Speaking at the forum, Song Dae-seop, NAVER AI Safety Policy leader, explained the rationale behind ASF 2.0, saying, “As AI technology, services, and the policy and regulatory landscape continue to evolve, the challenge is no longer simply how to make a single model safe, but how to safely design and operate multi-model services used by tens of millions of people.”
Song added that NAVER ASF 2.0 incorporates a range of considerations, including ▲NAVER’s ongoing development of its On-service AI strategy through services such as AI Tab and Shopping AI Agent; ▲the growing adoption of multi-model environments across the global AI ecosystem; and ▲shifts in the policy and regulatory landscape, such as the enactment of Korea’s AI Basic Act.
[Image] NAVER ASF 2.0
ASF 2.0 aims to manage AI safety from the perspectives of users and services, covering every stage of the AI service lifecycle, from launch through ongoing operation. To this end, ASF 2.0 expands safety management beyond individual AI models to services operating in multi-model environments. It also replaces a single performance-based criterion with separate criteria for context, use case and impact.
Specifically, the ▲AI Risk Taxonomy categorizes potential risks in AI services, while the ▲AI Impact Assessment Matrix assesses their anticipated impact based on the area and scope of use. ▲Finally, NAVER will use these systems to conduct ongoing safety assessments and draw on user feedback to help users continue to use AI services safely.
NAVER has also established CHEC 2.0, a company-wide implementation framework designed to ensure that ASF 2.0 is applied consistently throughout the service launch process. AI Tab, introduced in June, was also reviewed for AI safety through CHEC 2.0 at every stage, from design to launch. NAVER plans to continue using CHEC 2.0 to assess the safety of both its upcoming and already-launched AI-powered services, including AI Tab.
* CHEC 2.0: Consultation on Human-centered AI’s Ethics and Safety Considerations
Song said, “Since unveiling the NAVER AI Ethics Principles in 2021 and NAVER ASF Beta in 2024, NAVER has continued to refine its AI safety framework in step with global developments in AI ethics and Korea’s evolving policy landscape. Going forward, we will continue to deepen our experience and expertise in AI safety, working with academia, policymakers, external experts and a wide range of organizations to share and further develop that expertise.” (End)
