There is no basis anywhere to think it will not be targeted.

Place a product label as the condition for connecting to the grid, and keep unlabeled product groups out. Separate from whether this rule is right or wrong, it is necessary to consider how the grid after the rule is established will look to an attacker. Supply converges on specific product groups, those product groups carry the outward appearance of being "countermeasure-ready," and they also carry the political context of exclusion. If a compromise succeeds, in addition to technical damage such as a blackout, there will be the result that "the system that said it would protect us with labels was breached." These three conditions increase the attacker’s cost-effectiveness, and the draft revision satisfies all three at once. The threat model that the ministry answered in writing as "nonexistent" was precisely what was needed to deny this structure.

益田 周防海(Suomi Masuda)/ I-S3 Co., Ltd. | October 17, 2026

Attackers do not choose targets by motive. They choose by the ratio of the cost required to reach them, the gain if successful, and the burden if attributed. The draft revision of the grid connection technical requirements guideline (e-Gov case number 620340012) shifts this ratio in the attacker’s favor. Demand for new installations and replacements concentrates on the product groups that passed the label, and the capacity belonging to the same firmware distribution system and same control plane increases (the gain per cost of access increases). The label becomes an outward sign of being "countermeasure-ready," and boundary defenses and limits on reachable capacity on the installation side are more likely to be omitted (the cost of access goes down). And because it is a system described as excluding specific products, a successful compromise damages the credibility of the system itself (the gain exceeds technical damage). The most valuable target is the manufacturer’s update distribution server and signing key, and JC-STAR Star 1 does not evaluate that.

Explainer video (about 9 min)

Other language versions: Japanese / English / Simplified Chinese / playlist / First installment (the five items of underwriting review, about 6 and a half minutes): the threat of labels versus reality / How to submit comments: submit comments by November 6

1. Criteria an attacker uses to select a target

Publicly available cases of attacks on power equipment share a common pattern in how targets are chosen. Rather than taking down individual weakly defended devices one by one, they use a path that can reach many devices at once in a single access.

Caseroute reachedscope reached at the same time
Ukrainian power distribution grid (December 2015 and December 2016)From intrusion into an operations terminal to the control plane of distribution SCADA. In 2016, malware directly operated substation protocols (IEC 60870-5-104, etc.)In 2015, about 230,000 customers lost power. In 2016, a substation north of Kyiv
Viasat KA-SAT (February 24, 2022)Delivered destructive updates to modems from the satellite communications operator's ground management networkTens of thousands of modems in Europe. In Germany, about 5,800 Enercon wind turbines lost remote monitoring
Danish energy operator (May 2023)Vulnerabilities in the boundary device commonly used by operators (Zyxel firewalls)According to a SektorCERT report, 22 operators were compromised simultaneously, and some were forced into island operation (disconnection from the grid)
SolarWinds Orion (2020)Infiltrated the vendor's build environment for monitoring software and inserted malicious code into a legitimately signed updateDistributed simultaneously to about 18,000 organizations that applied the update

What the four cases have in common is that the individual strength of the devices was not the issue. What was compromised was the management plane, distribution plane, and boundary devices that many devices commonly depend on, and once access was gained, many moved at the same time. Product-level certification does not evaluate this path.

2. Three conditions that the revision proposal must satisfy

The revised proposal itself does not write country names in the text. But the shape of the system after the制度 is enforced can be read from the text.

  1. As supply converges and the capacity belonging to a single distribution system increases, if labels are placed on the grid interconnection conditions, demand for new installations and replacements will concentrate on the product groups that have obtained the label. Because the product groups that obtain it are limited, the capacity belonging to the same manufacturer's same firmware distribution system and same cloud control plane becomes larger than it would be without the system. The draft revision does not include a rule that places an upper limit on the capacity that can move at the same time in a single incident (Cyber Blast Radius). This point was submitted to the competent authority as Comment 6 and accepted under receipt number 620340012420402491.
  2. With an outward appearance of being "already addressed," site-side defenses are likely to be omitted In its response dated July 26, 2026, IPA clearly states that the label indicates functions built into the product, and does not indicate that the product has been safely installed and operated. However, if the operational practice is to judge whether connection is allowed based on the label, installers, contractors, and general transmission and distribution operators will come to think that "the label is there, so that is enough." Site-side measures such as boundary defense, limiting reachable capacity, and output floors (minimum operating points) will be omitted to the extent they are costly, because they are not written into the requirements. The cost of reaching the system goes down.
  3. With an exclusionary context attached, the success of an intrusion affects the credibility of the system In public debate, this revision is being described as a measure aimed at products from a specific country. The competent authority has not said so explicitly, but the way it is received cannot be changed. If an intrusion succeeds under this context, the consequence is not limited to technical damage such as outages or output fluctuations. The fact that "the system that was supposed to be removed for safety was breached" itself becomes a greater gain for a state-like attacker than an outage. The cost stays the same, and only the gain increases.
This is not a guess that "attackers would think that way." It is the result of applying the system shape created by the draft revision to the attacker’s selection criteria: cost of access, gain, and attribution burden. To deny it, materials are needed to show that one of the three conditions does not hold.

3. The most valuable target, and the place the label does not see

From the three conditions, the most cost-effective target is the manufacturer's update distribution server and the firmware signing key. The reason is simple: if you reach that point, the update can be delivered through the legitimate path to all devices that have obtained the label at the same time. Signature verification on the device side cannot stop a malicious update that has been signed with the correct key.

The JC-STAR Star 1 conformity criteria check whether a product has a function to receive secure updates. How the server on the distribution side is operated, where the signing key is stored and who can use it, how revocation and re-signing are handled when the key leaks, and whether updates are distributed to all units at once or in stages—these are not product functions, so they are not evaluated. The IPA response saying it does not indicate installation or operation refers to this boundary.

the point of reach from the attacker's perspectivethe scope that operates when reachedIs it within the scope of JC-STAR★1 evaluation?
the management screen of a single device (default password, etc.)that one unitscope (such as eliminating default passwords)
network boundary at the installation siteequipment at that siteout of scope (installation side)
Manufacturer cloud control side (remote output control)all units belonging to that control sideout of scope (operations side)
manufacturer's update distribution server / signing keyall units belonging to that distribution system, via the legitimate routeout of scope (operations side)

The lower you go in the table, the broader the scope that can be affected by a single reach, and the farther it is from label evaluation. The draft revision is designed to determine connection eligibility by looking only at the top row of the table.

4. Materials needed to say it will not be targeted

To deny this structure, at least the following three documents are needed. All of them were requested in the public inquiry our company sent to the Agency for Natural Resources and Energy in July 2026.

required materialsWhat must be shown to deny the structureResponse from the supervising authority (July 2026)
threat modelThe assumed attacker, access path, and scope of damage are specified, and the update distribution side and control side are includedno material was shown as the basis for making it a requirement
comparative evaluation with alternative measuresA comparison showing that product selection by label is more effective than boundary defense or limiting reachable capacity"There is no comparative material"
impact assessmentAn estimate of how the capacity that operates simultaneously in one incident changes in a system after label requirements are introducedis not shown. The comparison between IP and non-IP is "not from the perspective of requirements"

The original text of the response is in the response dataset (CC BY 4.0), and the timeline with dates and question numbers is compiled in the ledger of foreseeability. It can be confirmed from the competent authority's own documents that there was no material at the stage of system design for determining that it would not be targeted.

5. Measures that can be taken now

Changing the structure requires revising the制度, but there are measures that can be taken without waiting for the制度. I will write by position.

installer / power producer / contractor

manufacturer

supervising authority / general transmission and distribution utility

Reader

Public comments on the draft revision are being accepted until November 6, 2026. Submission method and draft wording are prepared. If you submit the points of this article as they are, narrowing them to two points—"placing an upper limit on the capacity that can move at the same time in a single incident" and "including the update distribution plane and control plane in the threat model"—will make it easier for the competent authority to respond.

What this document does not say

We are not saying that a specific country or business operator is planning an attack. We are not predicting when or how likely an attack is to occur. We are also not saying that JC-STAR Star 1 is ineffective. Eliminating default passwords and mandating update functions are meaningful against intrusion into a single device. We are not determining that the ministry or the system operator intended targeting. What we are stating is only the result of applying the system shape created by the draft revision to the attacker’s selection criteria—cost of access, gain, and attribution burden—and the fact that, according to the ministry’s response, no materials exist to deny that result. We do not address the presence or absence of vulnerabilities in individual products, individual connection refusals, or the outcome of lawsuits.

Source

益田 周防海(Suomi Masuda)/ I-S3 Co., Ltd. Based on public materials and the written responses received. This is not an evaluation of any specific product or business operator, nor is it advice on any individual contract or lawsuit.