Expanding the AI Agent Environments and Safety Standards of Government AI GENAI Across Society: Introducing "Agent Chat" and "GENAI Workbench," and Publishing the "GENAI Standards"

Government operations involve not only information gathering and document drafting, but also complex, specialized tasks such as multi-step workflows and advanced research and analysis based on large volumes of materials. However, simply using existing AI applications is not enough to fully leverage AI in these tasks. Government employees need an environment in which they can create and run AI agents tailored to their duties and objectives and continuously improve them through practical use. Utilizing AI agents in such an environment makes it possible to promote frontline-led operational reform, improve efficiency and quality, and develop AI-ready talent.

To address this, the Digital Agency is beginning to provide two types of AI agent environments through Government AI GENAI for approximately 180,000 government employees. Starting today, "Agent Chat" will be rolled out sequentially across ministries and agencies, with the introduction of "GENAI Workbench" targeted for around December 2026.

  • 1. Agent Chat: An environment for easily creating and running AI agents
  • 2. GENAI Workbench: An environment for building and running advanced AI agents

According to research conducted by the Digital Agency, deploying an advanced AI agent environment—comparable to advanced AI coding agents (AI that autonomously generates and modifies code, performs file operations, executes commands, and runs tests based on natural language instructions)—at a scale of approximately 180,000 users as a common government platform available not only to technical specialists, but also to general administrative staff for their daily work, marks the world’s first initiative of its kind.

Furthermore, to enable widespread use by local governments, private businesses, and other organizations, the Digital Agency plans to release both Agent Chat and GENAI Workbench around February 2027 as "GENAI OSS (Open Source Software) Ver. 2.0," which anyone can use, modify, and redistribute.

In addition, to ensure that government employees can use AI agents safely and appropriately, the Digital Agency has established and published the "Government AI GENAI: Standards on Governance, Security, and Safety ("GENAI Standards") (Japanese text)（PDF／3,905KB）"("hereinafter referred to as the GENAI Standards"), which set out fundamental principles and compliance requirements for relevant parties.
AI agents have more advanced and complex capabilities than conventional AI applications and can perform a wider range of tasks. Clear rules are therefore essential to enable government employees to use these tools proactively while ensuring security and governance. The GENAI Standards clearly define compliance requirements for "Users," "Administrators," and "Developers" according to their respective roles.

The GENAI Standards were developed with the cooperation of the AI Safety Institute (AISI), which is responsible for examining and promoting methodologies and standards for evaluating AI safety. We expect these standards to serve as a useful reference model for local governments, companies, and other organizations considering the introduction of AI agent environments, helping them achieve both safety and effective utilization.

Related Links