Name in original language
デジタル社会推進標準ガイドライン DS-920 行政の進化と革新のための生成AIの調達・利活用に係るガイドライン
Initiative overview
As AI-related technologies advance rapidly, and their utilization expands across the public and private sectors, the Japanese Government recognized the necessity to establish rules and frameworks to promote the proper procurement and utilization of generative AI within its administrative operations, while ensuring appropriate risk management. Accordingly, the “Guideline for Japanese Governments’ Procurements and Utilizations of Generative AI for the sake of Evolution and Innovation of Public Administration” was approved by the Council for the Promotion of a Digital Society Executive Board on 27 May 2025.
The Guideline was developed by referring to domestic and international initiatives, including the AI Guidelines for Business, the outcomes of the Hiroshima AI Process, and discussions in frameworks such as the OECD and the G7.
The Guideline applies to the national government staff and covers text-generating AI systems, including those incorporating large language models (LLMs), used as components of government information systems, excluding systems handling classified or national security–related information.
Under the Guideline, each ministry or agency should appoint a Chief AI Officer (CAIO) responsible for promoting utilization while ensuring AI governance, risk management, and, enhance AI literacy among government staff. Based on the “Model User Rule for Generative AI” the CAIO develops rules for users within each ministry or agency.
The Guideline also establishes a government-wide Advanced AI Utilization Advisory Board to provide advice on projects likely to be classified as high-risk, share best practices, and consider revisions of the Guideline. It introduces the “High-Risk AI Project Finder” assessing risks based on factors such as users, operations, confidential information or personal data, and staff judgment on outputs. Given rapid technological advancement, the Guideline is to be reviewed and revised in a timely manner, with planned revisions from fiscal year 2026 onward, including scope expansion, enhancements to procurement check sheets, updates reflecting institutional and regulatory developments, and revisions to the High-Risk AI Project Finder.
