Name in original language
AIセーフティ・インスティテュート(J-AISI)
Initiative overview
The AI Safety Institute (AISI) was established by the Japanese government to advance research, standards and evaluation methodologies for safe and trustworthy artificial intelligence. Hosted within Japan’s Information-technology Promotion Agency (IPA), the institute operates as a national hub for technical assessment of advanced AI systems and for coordination with international AI safety efforts. AISI develops evaluation frameworks and testing protocols that enable governments, companies and researchers to assess the reliability, security and societal impacts of AI models. Its activities include building safety benchmarks, conducting model evaluation studies, developing safety guidelines and organising working groups that bring together industry experts, researchers and regulators. The institute also plays an important role in international cooperation on AI governance, particularly through engagement with initiatives such as the Hiroshima AI Process and other G7-related activities.



























