English < Article List - DigitalToday

Search results for Frontier AI Lab

AI & Enterprise

KT unveils Korean culture-focused AI safety benchmark KSAFE-MM

KT unveiled KSAFE-MM, a multimodal AI safety benchmark developed with Korea University to assess model safety in Korean-language and Korean cultural settings. The dataset includes KSAFE-MM-G, which adapts global risks to Korean cultural context, and KSAFE-MM-C, which reflects local issues such as rental deposit fraud and the Dokdo dispute. The benchmark contains 14,135 samples and was used to test 12 models. KT also built an automated evaluation pipeline and published results on arXiv and Hugging Face.

AI & Enterprise

Cryptolab, KT to co-develop encrypted AI solution to automate medical insurance review

Cryptolab said on May 28 it signed a joint research and development agreement with KT to develop a medical multimodal agentic AI solution based on AI security technology. The companies aim to build a medical AI infrastructure that lets AI agents analyze and infer without exposing patient data externally by combining KT\'s security cloud infrastructure with Cryptolab\'s homomorphic encryption-based generative AI security middleware. They plan to develop a multimodal AI agent to support hospitals\' pre-review of insurance claims.