Energy Law and AI Decision Accountability Chains in Energy Regulation in Japan

1. Introduction

AI decision accountability chains in energy regulation in Japan refers to the legal and institutional framework for determining who is responsible when artificial intelligence (AI), automated algorithms, or machine-learning systems influence decisions concerning electricity generation, transmission, distribution, energy pricing, renewable-energy integration, nuclear safety, and energy-market supervision.

Japan's energy sector increasingly relies on digital technologies to forecast electricity demand, optimize grid operations, detect equipment failures, manage renewable-energy output, and identify market irregularities. These technologies can improve efficiency, but they also create a fundamental legal question:

When an AI-assisted energy decision causes harm, violates a regulatory requirement, or unfairly affects a consumer or energy company, who is legally accountable—the regulator, utility, AI developer, system operator, or human official who approved the decision?

Japanese law does not establish one universal legal regime specifically governing every AI decision in the energy sector. Instead, accountability must be understood through several overlapping areas of law:

Energy regulation and electricity-market supervision.

Administrative law and judicial review.

Nuclear safety and environmental protection.

Personal information protection and cybersecurity.

Civil liability, contractual responsibility, and product liability.

General AI governance and human oversight.

The central principle is that using AI does not automatically transfer legal responsibility from the organization or public authority making the decision to the algorithm or its developer. The applicable responsibility depends on the governing statute, the nature of the decision, the parties' respective roles, and the evidence of fault or legal violation.

2. Meaning of an AI decision accountability chain

An accountability chain is the sequence of people, organizations, technical systems, and legal duties connecting the development of an AI model to the final regulatory or operational decision.

1. AI developer or vendor

Designs the model, software, training process and technical documentation.

2. Energy company or system operator

Validates the model, supplies operational data and establishes permitted uses.

3. Human operator or decision-maker

Reviews recommendations, applies operational safeguards and escalates abnormal outputs.

4. Regulatory or operational decision

A grid dispatch, market intervention, licence decision, safety measure or consumer-facing action.

5. Audit, investigation and remedy

Logs, incident reports, administrative review, compensation and corrective action.

This chain is not necessarily a simple hierarchy. A utility may operate an AI system supplied by a third party, while a public regulator independently evaluates the utility's conduct. Several parties may therefore have different, overlapping duties.

A sound legal analysis distinguishes four questions:

Causal responsibility: Which system or action contributed to the outcome?

Operational responsibility: Who deployed, supervised, or acted on the AI output?

Legal responsibility: Which person or organization breached an applicable legal duty?

Remedial responsibility: Who must correct the problem, pay compensation, or reconsider the decision?

These questions are related but not identical. For example, a software defect may cause an incorrect prediction, but a utility could also be responsible if it deployed the model without testing or ignored a mandatory safety procedure.

3. Japanese legal framework governing accountability

3.1 Electricity Business Act

The Electricity Business Act (電気事業法, Denki Jigyō Hō) is a principal statute governing electricity businesses in Japan. It provides the legal foundation for business regulation, electricity supply arrangements, technical safety obligations, and government supervision.

The Act is relevant when AI systems are used for:

Electricity supply-demand balancing and grid operations.

Network tariff calculations and regulated business activities.

Electricity trading and market surveillance.

Monitoring the technical safety of electrical installations.

Supporting decisions about the reliability and continuity of supply.

The Electricity and Gas Market Surveillance Commission (EGC), established in 2015, monitors electricity and gas markets and has statutory powers that include collecting reports, conducting on-site inspections, making business-improvement recommendations, and undertaking specified mediation and arbitration functions.

emsc.meti.go.jp

+1

Accountability implication: an electricity company cannot ordinarily defend a failure to comply with an applicable legal obligation merely by saying that an algorithm made the decision. The company must examine whether its own deployment, supervision, and response procedures complied with the relevant legal requirements.

For example, suppose a utility deploys an AI system that predicts demand incorrectly and automatically recommends reducing reserve capacity. If the resulting operational decision breaches an applicable supply or technical-safety requirement, investigators should examine both the model's performance and the utility's decision-making controls.

The legal question is not simply whether the AI made an error. It is whether a regulated entity breached a duty imposed by law, regulation, licence conditions, technical standards, or another applicable legal instrument.

3.2 Nuclear regulation and the Nuclear Regulation Authority

The Nuclear Regulation Authority (NRA) oversees nuclear safety under Japan's nuclear regulatory framework, including the Reactor Regulation Act and associated technical requirements.

AI may support nuclear operations through:

Predictive maintenance.

Detection of abnormal operating conditions.

Analysis of sensor data.

Radiation monitoring.

Identification of equipment degradation.

Simulation of accident scenarios.

However, an AI prediction is not a substitute for the legally required safety assessment, licensing process, or regulatory compliance.

An operator should be able to demonstrate what information informed a safety decision, which validated procedures applied, whether uncertainty was considered, and who had authority to approve or reject the AI recommendation.

This is particularly important where a model's prediction conflicts with a safety procedure or conventional engineering evidence.

3.3 Japan's AI governance guidelines

Japan's AI governance approach also informs accountability arrangements.

The Ministry of Economy, Trade and Industry (METI) and the Ministry of Internal Affairs and Communications (MIC) developed the AI Guidelines for Business. METI's published revision history records version 1.2 on 31 March 2026. The guidelines provide a cross-sector framework for AI development, provision, and use.

Ministry of Economy, Trade and Industry

+1

The guidelines are important for energy companies because they support governance measures such as:

Clearly assigning responsibility for AI systems.

Assessing risks before deployment.

Testing systems for reliability and limitations.

Monitoring performance after deployment.

Maintaining appropriate human oversight.

Providing information to relevant stakeholders.

Responding to incidents and correcting failures.

These guidelines should not automatically be treated as binding statutory rules. Their legal significance depends on the relevant provision, any applicable sector-specific requirement, contractual incorporation, and how they may inform the standard of reasonable conduct in a particular dispute.

Japan's AI-related legislation and policy framework should therefore be distinguished from the binding duties imposed by the Electricity Business Act, nuclear legislation, civil law, and other applicable statutes.

4. Constitutional constraints on AI-assisted energy decisions

The Japanese Constitution does not contain a specific article devoted to AI accountability in energy regulation. Nevertheless, constitutional principles can constrain public authorities that use AI to exercise regulatory powers.

4.1 Article 13: Individual rights and personal dignity

Article 13 protects respect for individuals and their right to life, liberty, and the pursuit of happiness, subject to the public welfare.

In energy regulation, this principle may be relevant when an AI-assisted decision threatens personal safety, privacy, or important individual interests.

Examples include:

An AI-supported decision that increases risks to people living near a nuclear facility.

Automated processing of consumer information for electricity-related services.

An algorithm that contributes to an unjustified interruption of essential electricity services.

Article 13 does not automatically establish a right to challenge every AI output. A claimant must identify the relevant legal interest and establish the necessary grounds for judicial or other legal relief.

4.2 Article 14: Equality before the law

Article 14 guarantees equality before the law and prohibits discrimination on the grounds specified in the constitutional provision.

An AI model may create equality concerns if it systematically disadvantages particular consumers, regions, or classes of regulated businesses.

For example, an automated risk-scoring system might classify a group of small electricity suppliers as unusually risky because its training data disproportionately reflects the historical failures of similar businesses.

If a public authority uses that score in a regulatory decision, the relevant questions include:

Was the classification based on legally relevant criteria?

Did the authority examine whether the underlying data were reliable?

Were comparable entities treated consistently?

Did the authority exercise its statutory discretion independently?

Article 14 does not mean that every difference in AI-generated outcomes is unconstitutional. The legal assessment depends on the nature of the classification, the governmental action, and the applicable standards of review.

4.3 Article 31: Due process and legal procedures

Article 31 expressly concerns deprivation of life or liberty through criminal procedure. Its application to administrative proceedings is not identical to its application in criminal cases.

The Supreme Court's 1992 Ikata Nuclear Power Plant judgment addressed this distinction. The Court explained that administrative procedures differ according to their nature and purpose, and that procedures such as prior notice and an opportunity to be heard are not automatically required in every administrative proceeding.

Supreme Court of Japan

For AI-assisted regulation, this means there is no universal constitutional rule requiring a hearing whenever an algorithm is used. The applicable legislation, the nature of the decision, and the affected person's rights must be examined.

Where a statute requires notice, reasons, consultation, or a hearing, using AI cannot excuse the authority from following that procedure.

4.4 Article 32: Access to courts

Article 32 provides that no person shall be denied the right of access to the courts.

Where an AI-assisted administrative decision is legally reviewable, the use of an algorithm should not itself eliminate access to judicial remedies.

For instance, a company challenging a regulatory decision should be able to contest the legality of the decision through the applicable procedure. Whether it can obtain source code, internal model documentation, or particular technical records depends on the relevant procedural and disclosure rules; Article 32 does not automatically guarantee unrestricted access to every technical document.

4.5 Article 76: Judicial independence

Article 76 vests judicial power in the courts and protects judicial independence.

An AI system may help regulators assess technical evidence, but its conclusions cannot displace the courts' legal responsibility to decide disputes within their jurisdiction.

A court reviewing a nuclear safety decision, for example, must apply the relevant legal standards to the evidence and administrative process. It should not treat an AI-generated safety score as conclusive merely because it was produced by a sophisticated model.

5. Allocation of accountability among the participants

The accountability chain becomes legally useful when each participant's role is examined separately.

These are questions of potential responsibility, not automatic findings of liability.

A vendor may be liable under a contract or an applicable liability rule, while the utility may independently be responsible for negligent deployment. A regulator may be responsible for an unlawful administrative decision, but not necessarily for every loss that follows it.

The same incident can therefore generate several legal claims, each with its own elements, defences, and causation requirements.

6. Important Japanese case laws relevant to AI decision accountability

Important qualification: the following judgments are relevant by analogy to AI accountability. They are not judgments establishing a comprehensive, AI-specific liability regime for energy regulation. Their value lies in the principles they establish concerning technical evidence, regulatory discretion, causation, and judicial review.

Case 1: Ikata Nuclear Power Plant Licensing Case — Supreme Court, 29 October 1992

Case identification

Supreme Court of Japan · First Petty Bench · 29 October 1992

Legal issue: How should a court review an administrative decision based on highly technical nuclear-safety assessments?

Decision: The Supreme Court held that judicial review should examine whether the technical assessment underlying the administrative decision contains unreasonable aspects. It identified unreasonable examination criteria and material errors or omissions in the technical investigation, deliberation, or assessment process as grounds that can make the resulting licensing decision unlawful.

Supreme Court of Japan

Relevance to AI accountability

Imagine that the regulator uses a machine-learning model to assess whether a nuclear installation satisfies safety requirements.

The model may have been trained on incomplete operational data or may fail to account for an important class of equipment failures. If the regulator relies on that model without adequately examining its limitations, the reasoning in Ikata provides a framework for questioning the lawfulness of the resulting decision.

The court's focus is not simply on whether a sophisticated technical instrument was used. It is on the reasonableness of the assessment and the integrity of the process on which the administrative decision relies.

The judgment also discusses the burden of presenting evidence. Although the claimant generally bears the initial burden of alleging and proving unreasonable aspects, the administrative authority must first present substantial evidence and materials supporting the criteria and assessment process on which it relied. If it fails to do so, the court may factually presume that the assessment contains unreasonable aspects.

Supreme Court of Japan

Principle for AI governance: a regulator should retain the technical evidence needed to explain why an AI-assisted assessment was considered reliable and legally sufficient.

The case does not establish that every AI model must be fully transparent or that every model error invalidates a licence.

Case 2: Monju Nuclear Reactor Licensing Litigation — Supreme Court, 30 May 2005

Court: Supreme Court of Japan, First Petty Bench

Subject: The legality of the installation approval for the Monju prototype fast breeder reactor.

Outcome: The Supreme Court overturned the Nagoya High Court, Kanazawa Branch's judgment that had invalidated the approval and rejected the residents' appeal, leaving the government's victory in place.

ATOMICA -

+1

The litigation involved the adequacy of nuclear-safety review, including technical questions concerning sodium leakage and reactor accident scenarios. The Supreme Court concluded that the safety-review process did not contain the kind of serious, material errors or omissions necessary to invalidate the challenged approval.

ATOMICA -

Relevance to AI accountability

Monju illustrates the importance of distinguishing a technically controversial event from proof that an earlier administrative approval was unlawful.

In an AI-assisted energy system, a model's subsequent failure does not automatically establish that the original deployment or regulatory approval was unlawful. A court would need to consider the information reasonably available at the relevant time, the applicable safety criteria, the adequacy of the review, and the legal significance of any defect.

For example, if a predictive-maintenance model fails to detect a reactor component defect, the investigation should examine whether the model was reasonably validated against known failure modes and whether the responsible operator followed the applicable safety requirements.

The judgment is not an endorsement of AI-based decision-making. It demonstrates the importance of evaluating the actual technical assessment and the legal standard applicable to the challenged decision.

Case 3: Fukushima nuclear accident State Redress litigation — Supreme Court, 17 June 2022

Official case reference

Supreme Court · Second Petty Bench · Case No. 2021 (Ju) 342

Legal issue: Whether the State was liable under Article 1(1) of the State Redress Act because the Minister of Economy, Trade and Industry had failed to exercise regulatory authority to prevent the Fukushima Daiichi accident.

Decision: The Supreme Court rejected the relevant claims against the State on the grounds examined in its judgment, particularly the failure to establish the necessary causal relationship between the alleged regulatory omission and the accident.

Supreme Court of Japan

+1

The judgment is particularly important because it separates two questions:

Whether a public authority failed to exercise regulatory power lawfully.

Whether that failure had the necessary causal relationship with the damage for which compensation is claimed.

The Court concluded that the evidence did not establish that exercising the relevant regulatory authority on the assumed basis would have prevented the accident or a similar accident under the circumstances considered.

Supreme Court of Japan

Relevance to AI accountability

Suppose a government regulator uses an AI forecasting system to identify tsunami risks to energy infrastructure. The system produces an inadequate forecast, and the regulator does not require the operator to implement additional protective measures.

A later claim against the State would not necessarily succeed merely because the forecast was inaccurate or because the regulator did not act. The claimant would need to establish the elements of the applicable legal cause of action, including the relevant unlawfulness and causation requirements.

Likewise, the fact that a regulator used AI would not automatically immunize it from liability. The legal analysis would still examine the regulator's statutory authority, the information available, the nature of the omission, and the connection between that omission and the loss.

The 2022 judgment therefore provides a particularly important framework for distinguishing technical error, regulatory fault, and compensable damage.

Case 4: Ohi Nuclear Power Plant injunction judgment — Fukui District Court, 21 May 2014

Court: Fukui District Court

Subject: Residents' request for an injunction against the operation of Units 3 and 4 of the Ohi Nuclear Power Plant.

Outcome: The court granted an injunction for the plaintiffs covered by its ruling, relying on the protection of personal rights and its assessment of the risks associated with nuclear power plant operation.

Réseau Sortir du nucléaire

+1

This judgment is significant for the role of courts in assessing serious risks to life, health, and livelihood. It is a lower-court decision and should not be presented as a universal or final statement of Japanese nuclear-safety law.

Relevance to AI accountability

Consider a situation in which an AI-based safety model estimates that a nuclear plant's risk is acceptably low, while residents produce evidence suggesting that the model fails to account for a significant hazard.

The Ohi litigation illustrates that safety-related legal disputes can require courts to assess evidence about potential harm and personal rights, rather than treating an operator's technical assurance as conclusive.

It does not establish that an AI prediction alone can justify an injunction. The legal requirements for an injunction, the available evidence, and the applicable procedural standard remain essential.

Case 5: Ikata Nuclear Power Plant injunction proceedings — Hiroshima High Court, 17 January 2020

The Hiroshima High Court issued a preliminary injunction concerning operation of the Ikata Nuclear Power Plant. The ruling was part of a series of post-Fukushima disputes in which courts assessed claims about nuclear-safety risks and the adequacy of relevant assessments.

OECD

The decision is relevant to AI-supported risk analysis because it demonstrates that the evaluation of technical risk can have immediate legal consequences for an energy operator.

If a regulator or operator relies on a predictive model to assess volcanic hazards, seismic activity, or other natural risks, an affected party may contest the assumptions, evidence, or risk assessment through the applicable legal procedure.

The 2020 decision should be read in its procedural and factual context. Later decisions in the Ikata proceedings reached different conclusions, illustrating that the outcome depends on the evidence and legal issues before the court.

裁判所判例Watch

+1

What these cases establish collectively

The important doctrinal distinction is that these cases provide analogical principles, not a judicially created rule that every AI decision must be explainable in a particular technical format.

7. Civil liability for AI errors in Japan's energy sector

Accountability also depends on whether an AI-related incident gives rise to a claim under Japanese civil law.

7.1 Civil Code Article 709: Tort liability

Article 709 of the Japanese Civil Code establishes liability for loss or damage caused by an intentional or negligent infringement of another person's rights or legally protected interests.

Potential applications include:

A utility negligently deploying an inadequately tested AI control system.

A developer failing to exercise appropriate care in designing a safety-critical model.

An operator ignoring warnings that indicate an imminent equipment failure.

A company using AI-generated recommendations in a way that unlawfully harms consumers or competing businesses.

A claimant must establish the elements of the applicable tort claim. These generally include an infringement of a right or legally protected interest, intent or negligence, damage, and causation.

Example: An AI system fails to predict overheating at a substation. If the utility failed to implement reasonable validation and monitoring procedures and the failure caused foreseeable property damage, Article 709 may be relevant.

However, the fact that the model produced an inaccurate result does not automatically establish negligence. The circumstances of development and deployment, the applicable standards, and the parties' conduct matter.

7.2 Civil Code Article 415: Liability for breach of contract

Article 415 concerns liability for non-performance of contractual obligations.

It may apply where an AI vendor has agreed to provide a system meeting specified performance, security, maintenance, or reporting requirements but fails to fulfil those obligations.

For instance, an energy company may contract with an AI supplier to provide a demand-forecasting system with specified accuracy levels and incident-response support. If the supplier breaches an applicable contractual obligation, the company may have a contractual claim, subject to the requirements and defences under Japanese law.

Contractual allocation of risk does not automatically eliminate statutory duties owed to third parties or public authorities.

7.3 Product Liability Act

Japan's Product Liability Act may become relevant when a defect in a covered manufactured movable product causes death, injury, or damage to other property.

Its application to AI-related energy incidents depends on the product involved and the nature of the defect. For example, a defective physical controller incorporating AI could raise different issues from a stand-alone software service.

It should not be assumed that every software malfunction falls within the Act. Contract law, tort law, and other applicable rules may be more relevant in a particular case.

7.4 State Redress Act, Article 1

Where a public employee unlawfully causes damage intentionally or negligently in the course of exercising public authority, Article 1(1) of the State Redress Act may provide a basis for compensation by the State or relevant public entity.

If an AI-assisted regulatory decision is alleged to be unlawful, the inquiry includes the official conduct, the applicable statutory duties, and causation. A public authority is not automatically liable simply because its AI system produced an incorrect output.

The 2022 Fukushima Supreme Court judgment demonstrates the importance of the specific legal requirements for State liability and the causal relationship between the alleged omission and the damage.

Supreme Court of Japan

8. Practical accountability mechanisms for energy regulators

The following are governance recommendations for applying Japan's existing legal framework. They should not be mistaken for a claim that Japanese law already prescribes every mechanism as a universal statutory requirement.

A. Prior authorization and risk classification

Before deployment, classify the AI system according to its impact. A model that generates internal demand forecasts requires different controls from one that directly controls grid equipment or supports nuclear-safety decisions.

B. Named responsibility

Identify the accountable utility executive, system owner, technical operator, and relevant vendor contacts. Define who may approve model changes and who must respond to unsafe outputs.

C. Audit trails

Preserve relevant inputs, model versions, recommendations, uncertainty estimates, human overrides, approvals, and resulting actions. Records should be sufficient to reconstruct important decisions and investigate incidents.

D. Independent validation and monitoring

Test the system against unusual operating conditions, data errors, cyberattacks, and foreseeable failure modes. Reassess it after material changes in equipment, data, or operating conditions.

E. Review and remedies

Provide appropriate escalation procedures, incident investigation, correction of unlawful decisions, and access to applicable administrative or judicial remedies.

These mechanisms align with the general direction of Japan's AI governance guidance and the evidence-based regulatory approach reflected in its energy and nuclear case law.

Ministry of Economy, Trade and Industry

+2

9. Hypothetical applications in Japan

The following scenarios illustrate how an accountability chain could be analysed. They are hypothetical examples, not reported Japanese court decisions.

Scenario A: AI-assisted electricity dispatch

An AI system forecasts a shortage of renewable electricity and recommends reducing output from a particular generator while increasing output from another.

The affected generator claims the decision was discriminatory.

The investigation should establish:

Who designed and operated the algorithm?

Which market rules and operational constraints governed dispatch?

Were the input data accurate and sufficiently complete?

Did the operator follow the applicable market rules?

Was the decision based on legitimate operational considerations?

Is there evidence of a breach, measurable loss, and legally sufficient causation?

The EGC's market-monitoring and supervisory functions may be relevant, depending on the issue and its statutory powers.

emsc.meti.go.jp

+1

Scenario B: AI fails to detect a nuclear safety hazard

A predictive-maintenance model identifies a reactor component as operating normally, despite signs of degradation. A subsequent failure causes damage.

Potentially relevant parties include the model developer, system integrator, nuclear operator, maintenance contractor, and responsible personnel.

The investigation would examine whether the model was suitable for the intended purpose, whether known limitations were communicated, whether mandatory inspections were carried out, and whether warning signs were ignored.

The Ikata licensing judgment provides a framework for examining the adequacy of a technical assessment underlying a regulatory decision. The Fukushima State Redress judgment separately illustrates the importance of causation when State liability is claimed.

Supreme Court of Japan

+1

Scenario C: AI-assisted licence refusal

A public authority uses an AI model to assess applications for an energy-related permit. The model incorrectly classifies one applicant as high-risk, and the authority refuses the application.

The principal questions are whether the authority acted within its statutory powers, followed applicable procedures, considered relevant evidence, and exercised its discretion lawfully.

If the applicant challenges the refusal, the administrative record and the applicable judicial-review rules will be important. Depending on the case, remedies may include seeking revocation of the administrative disposition or other relief available under the governing law.

The authority cannot assume that an algorithm's recommendation conclusively establishes the legality of its decision.

10. Recommended model for an AI accountability chain

For a Japanese energy company or regulator, a useful governance model combines six elements.

The intensity of controls should be proportionate to the potential harm. AI used for routine internal reporting does not necessarily need the same governance arrangements as AI that directly influences grid protection or nuclear-safety decisions.

Japan's EGC guidance and the AI Guidelines for Business are useful starting points for developing this model, while the applicable legislation determines which duties are legally mandatory in each situation.

emsc.meti.go.jp

+1

11. Critical legal challenges

Several unresolved or context-dependent issues deserve particular attention.

First, opacity. Machine-learning models may be difficult to explain. An authority must nevertheless be able to establish a legally adequate basis for a decision where the governing law requires it.

Second, fragmented responsibility. The developer, vendor, utility, and regulator may control different parts of the decision chain. Contracts and internal policies can allocate tasks, but they do not necessarily determine liability to third parties.

Third, causation. Energy systems are complex. A model error may contribute to an incident without being its sole cause. Legal responsibility requires analysis under the applicable causation standard.

Fourth, technical uncertainty. A model may have been reasonable when deployed but become unreliable after changes in equipment, climate patterns, or operating conditions. Ongoing monitoring is therefore an important governance safeguard.

Fifth, public accountability. A public authority must exercise the legal powers assigned to it. It cannot treat an algorithm as a substitute for compliance with governing statutes, required procedures, or judicial review.

12. Conclusion

AI decision accountability chains in Japanese energy regulation are best understood as a combination of energy-sector obligations, administrative-law controls, civil liability, constitutional principles, and general AI governance.

The key judicial authorities are the Ikata Nuclear Power Plant Licensing Case (1992), the Monju Nuclear Reactor Licensing Litigation (2005), the Fukushima State Redress judgment (2022), the Ohi injunction judgment (2014), and the Ikata injunction proceedings (2020). They address different questions and must not be treated as a single body of AI-specific precedent.

Taken together, these authorities help explain why an AI-assisted energy decision should be supported by a defensible technical assessment, why public authorities must comply with their legal duties, why courts can examine the basis of technical decisions, and why compensation depends on the applicable liability requirements.

The central legal conclusion is that AI can support energy decision-making, but it does not eliminate the need to identify the legally responsible decision-maker, establish the applicable duty, preserve evidence, and provide appropriate remedies.

Selected primary and official sources

Supreme Court of Japan — Ikata Nuclear Power Plant Licensing Case, 29 October 1992.

Supreme Court of Japan — Fukushima State Redress judgment, 17 June 2022.

Electricity and Gas Market Surveillance Commission — Official English website.

emsc.meti.go.jp

+1

Ministry of Economy, Trade and Industry — AI Guidelines for Business, version history and official materials.

Ministry of Economy, Trade and Industry

Japan Atomic Energy Agency, ATOMICA — History of the Monju litigation and Supreme Court judgment.

ATOMICA -

Academic note: For a formal dissertation, article, or research paper, the Japanese-language judgments and the relevant statutory text should be consulted for precise quotations, procedural history, and the current wording of the legislation.