Naver on Tuesday unveiled ASF 2.0, an artificial intelligence (AI) safety framework that expands the scope of safety checks beyond existing technology model-focused AI risk management to cover services as actually experienced by users.

ASF 2.0 is an upgraded version of a framework announced at the 2024 AI Seoul Summit, refined to align with the latest global trends. Naver attended the AI Safety Seoul Forum held on Tuesday in Seoul's Gangnam district and shared the detailed components of a new safety management system reflecting policy and institutional changes, including the spread of multi-model environments and enactment of the AI Basic Act.

ASF 2.0 aims to manage the full lifecycle, from service launch to overall operations, by specifying evaluation criteria that focus on a single performance metric into context, use cases and impact. It operates a system of ongoing evaluation and feedback by applying an "AI risk classification system" that categorises risks that can arise in services and an "AI impact assessment matrix" that diagnoses expectations by application area.

Song Dae-seop (송대섭), Naver's AI Safety Policy leader, said changes in the technology, service, policy and institutional environment around AI are expanding the issue beyond making a single model safe to how to safely design and operate services used by tens of millions of people by combining various AI models. He explained the background to the initiative.

The company also built "CHEC 2.0," a companywide execution system, so ASF 2.0 can be applied consistently during the service launch process. Naver's "AI tab," introduced in June, also completed safety verification from design through launch under the system, and the company plans to continue checking AI-based products planned for release or already in service.

Song said Naver has been making its AI safety framework more concrete to fit the domestic policy environment by releasing the Naver AI Ethics Guidelines in 2021 and the Naver ASF beta in 2024. He said it will continue to build know-how and share and develop such experience through collaboration with academia, policy circles, external experts and various institutions.