Definition
The AI Business Operator Guidelines integrate and update prior "AI Ethics Guidelines" and "Human-Centric AI Society Principles" into a single soft-law document. Co-authored by METI and MIC; revised on a rolling basis (Version 1.x).
Ten principles: (1) human-centric, (2) safety, (3) fairness, (4) privacy, (5) security assurance, (6) transparency, (7) accountability, (8) education and literacy, (9) fair competition, (10) innovation. Each principle is mapped to the developer / provider / deployer roles.
Not legally binding, but the de facto reference for government procurement standards and industry self-regulation in Japan. Covers similar ground to the EU AI Act's high-risk requirements while adopting a technology-neutral, risk-based approach.
Lemma compliance path
Lemma provides the technical substrate for the transparency, explainability, and auditability the guidelines require. docHash + commitments + zero-knowledge proofs turn AI governance reports into concrete evidence.
AI input, model, and output are retained as an audit trail; when accountability is triggered, only the necessary attributes are released through selective disclosure.
Run in parallel with EU AI Act compliance and a globally-operating Japanese company can unify its AI governance across both regimes.