AI Regulation Tracker / National strategy
Japan Adopts Phase II AI Basic Plan Under the AI Promotion Act
On July 14, 2026, Japan's Cabinet adopted the Phase II AI Basic Plan (人工知能基本計画（第Ⅱ期）), a non-binding national strategy that sits under the AI Promotion Act. It points at agentic systems, physical and vertical AI, and building domestic AI capacity.
Japan's Cabinet adopted its Phase II AI Basic Plan on July 14, setting national priorities around agentic AI, vertical and physical AI, domestic AI capabilities, governance, and stronger model-evaluation capacity at Japan's AI Safety Institute.
What the plan actually does
Start with what this is not. It is not a new statute, and it is not a penalty regime. It is a strategy document adopted by Cabinet decision, and it sits underneath the AI Promotion Act, the promotion-style law that already frames how Japan handles AI. If you want the mechanics of that statute, we cover it separately in our Japan AI Promotion Act briefing. This piece is about the plan, which is the government telling ministries, industry, and research institutions where to point.
It is also Phase II. Japan's first AI Basic Plan was decided by Cabinet on December 23, 2025. The July 2026 instrument is the second one, the next turn of the same cycle, so read it as continuation rather than a standing start.
The emphasis in the final plan is fairly clear. It commits to annual revision for the time being, with monitoring against appropriate benchmarks, which is the part operators should note, because a plan that gets rewritten every year with numbers attached is a plan the government intends to hold itself to. It leans into agentic AI, into vertical and physical AI, and into strategic domestic AI capacity and sovereignty, meaning Japan wants more of the stack built and controlled at home. On the governance side it calls for stronger model evaluation, traceability, and guardrails at Japan's AI Safety Institute, and it keeps the door open on institutional and legal arrangements through a continuing review.
Who this affects
If you run or advise a company operating in Japan or selling into it, this is a planning input. The plan does not create a new obligation you have to comply with tomorrow. What it does is tell you which way the wind is blowing: toward more capable evaluation of frontier models, more attention to how AI systems can be traced and audited, and more state interest in domestic capacity. Ministries carry the direct workload, since they have to translate the priorities into programs and budgets. Compliance and strategy leads are the ones who have to explain to leadership why a non-binding document still deserves a slot on the risk map.
So what, for advisors and executives with Japan exposure
Keep the framing honest with your stakeholders. This is agenda-setting. Nobody gets fined under a Cabinet strategy. The reason to watch it is that the strategy names the areas where binding rules tend to show up later, and the AI Safety Institute language plus the standing review of legal arrangements are the tells. If your Japan footprint touches frontier or agentic systems, the practical move is to start tracking how AISI's evaluation and traceability expectations firm up, because that is where a soft priority can harden into a real requirement over time.
Questions professionals are asking
Is this Japan's first AI Basic Plan?
No. Japan's first AI Basic Plan was adopted by Cabinet decision on December 23, 2025. The instrument adopted on July 14, 2026 is Phase II, effectively the second plan in the same cycle.
Is the Phase II plan legally binding? Are there fines?
No. It is a non-binding national strategy adopted by Cabinet decision under the AI Promotion Act. It sets priorities and direction and does not create penalties.
How is this different from the AI Promotion Act?
The AI Promotion Act is the statute. The Phase II AI Basic Plan is a strategy adopted under that statute. The plan sets national direction, while the Act is the legal framework, and we cover the Act separately.
What should firms with Japan exposure watch next?
The annual revision for the time being, with monitoring against appropriate benchmarks, the strengthening of model evaluation and traceability at Japan's AI Safety Institute, and the continuing review of institutional and legal arrangements.
RELATED BRIEFINGS
Browse the full AI Regulation News tracker
Informational analysis for working professionals, not legal advice. Confirm how any rule applies to your situation with qualified counsel in the relevant jurisdiction.