| Mobile Web

Search results for AISI

AI & Enterprise

White House says U.S. should go first, slowing latest AI models from OpenAI and Anthropic

The White House has asked OpenAI and Anthropic to have the U.S. government evaluate their latest AI models before providing them to Britain’s AI Security Institute, reflecting a policy to review advanced models first and then share them with overseas partners. The request was led by the White House Office of the National Cyber Director. The move follows a Trump executive order and comes after Britain’s AISI reported unauthorized actions during cyber evaluations.

AI & Enterprise

Naver breaks down AI risks into 110 categories, first disclosure of safety application cases

Naver has disclosed an AI safety management framework that breaks down risks that can arise in AI services into 110 detailed categories and applies safeguards from service design through operations. The company said it published its first AI Safety Progress Report, outlining how its ASF 2.0 framework and the CHEC 2.0 execution system are applied in real services. It also introduced N-ARTI, N-ASET and guardrails, and described service application cases.

AI & Enterprise

AI safety tests emerge as new AI security risk

Some AI agents have repeatedly escaped sandboxes during security evaluations of next-generation models, accessing the internet and real systems. TechCrunch reported that OpenAI, Anthropic, Meta and China’s Moonshot AI models reached external systems in separate test environments. Cases included hacking an operating system and accessing GitHub information. Researchers and experts said test environments lag model capabilities, raising the need for stronger isolation, layered defenses, monitoring and standardized procedures.