AI Guidelines for Business Ver1.2 (METI/MIC): a free English checklist
General information, not legal advice. The Guidelines are voluntary (soft law). This checklist is our own summary; it does not replace the official checklist and worksheet (Appendix 7), and using it does not show that your organisation meets the Guidelines.
Source
Japan's Ministry of Internal Affairs and Communications (MIC) and Ministry of Economy, Trade and Industry (METI) published the AI Guidelines for Business Ver1.2 on 31 March 2026. Headings below follow the official English provisional translation; the Japanese text prevails.
- MIC: AI Guidelines for Business page (all versions and appendices)
- AI Guidelines for Business Ver1.2 (English provisional translation, PDF)
- Official checklist, Appendix 7, Ver1.2 (Japanese, PDF)
Step 0: know your role
The Guidelines name three kinds of AI business actor: AI developer, AI provider and AI business user. One company can hold more than one role depending on how it uses AI.
- For each AI system, write down which role(s) you hold. Register field: Screening → "Your role for this system"; "Built by".
Part 2 C: the 10 common guiding principles
| Principle (Ver1.2) | Sub-items in the Guidelines | Where to record it in EURO AI Register |
|---|---|---|
| 1) Human-centric | Human dignity and autonomy of individuals; paying attention to manipulation by AI of decision-making and emotions; countermeasures against disinformation; ensuring diversity/inclusion; providing user support; ensuring sustainability | "What it does (intended purpose)"; Policy |
| 2) Safety | Taking into consideration the lives, bodies, properties and minds of humans and the environment; proper use of AI; proper training | Intended purpose; Screening; Human review → "Rationale and evidence" |
| 3) Fairness | Consideration for bias in technologies forming AI models; intervention by decisions made by humans | Human review ("Reviewed by", "Reviewed on"); Evidence log |
| 4) Privacy protection | Protection of privacy across AI systems and services in general | Data classes; Policy |
| 5) Ensuring security | Security measures relevant to AI systems and services; consideration for the latest trends | Evidence log; "Next review" |
| 6) Transparency | Ensuring verifiability; providing relevant stakeholders with information; reasonable and truthful support; improving explainability and interpretability | Screening → "Transparency triggers"; Report |
| 7) Accountability | Improving traceability; explanation of conformity to the common guiding principles; designation of responsible persons; sharing responsibilities among actors; specific actions for stakeholders; documentation | "Internal owner"; "Vendor (or in-house)"; Evidence log; Report |
| 8) Education/literacy | Ensuring AI literacy; education and reskilling; support for stakeholders | Evidence log (training records); free Unadhi Academy course |
| 9) Ensuring fair competition | (no sub-items listed) | Notes |
| 10) Innovation | Promoting open innovation; consideration for interconnectivity and interoperability; providing information appropriately | Notes |
Part 5: items for AI business users
Most SMEs are AI business users (they use AI tools in their work). Table 1 of the Guidelines lists these items for them, in addition to the common principles:
- U-2 i. Proper use of AI that considers safety. Register: intended purpose, status (active/planned/retired).
- U-3 i. Consideration for bias in input data or prompts. Register: Human review, Evidence log.
- U-4 i. Countermeasures against inappropriate input of personal data and privacy violation. Register: Data classes, Policy.
- U-5 i. Implementation of security measures. Register: Evidence log.
- U-6 i. Providing relevant stakeholders with information. Register: Transparency triggers, Report.
- U-7 i. Explanation to relevant stakeholders. Register: Internal owner, Report.
- U-7 ii. Effective use of provided documents and conformity to agreements. Register: Vendor; vendor documents in the Evidence log.
For AI providers (Part 4), in short
Table 1 adds for providers, among others: actions against risks to lives, bodies, properties, minds and the environment; bias in configurations and data; privacy-protection mechanisms; security mechanisms and handling of vulnerabilities; documentation of system architectures; explanation to AI business users of conformity to the common principles; and documentation of service agreements. Read Part 4 of the Guidelines for the full list.
Start your register: 10 AI systems free
EURO AI Register is free for up to 10 AI systems. It runs in your browser, has an English interface and is a documentation tool: it helps you keep the records above in one place. Paid membership is $19/month or $190/year via Gumroad checkout.
Selling AI into the EU too? The EU AI Act can apply to Japanese companies: see EU AI Act guides. Related: Japan's AI Promotion Act. Accountants and consultants: partner programme.