[Report][SCSK Security Corporation] Learning from Japanese Cases and International Standards: Risks and Governance of Frontier AI - CODE BLUE 2026 #codeblue_jp #codeblue2026

[Report][SCSK Security Corporation] Learning from Japanese Cases and International Standards: Risks and Governance of Frontier AI - CODE BLUE 2026 #codeblue_jp #codeblue2026

A report on the session "[SCSK Security Corporation] Frontier AI Risks and Governance: Lessons from Japanese Cases and International Standards," held at CODE BLUE 2026.
2026.11.17

This page has been translated by machine translation. View original

Hello, this is Usuda.

Everyone, are you doing your security measures? (greeting

This is a report on the following session held at CODE BLUE 2026.

[SCSK Security Corporation] Learning from Japanese Cases and International Standards: Risks and Governance of Frontier AI

In 2026, the threat of frontier AI became a hot topic, but in the end what needs to be done is no different from before; the key point is how quickly it can be done.

That is not the whole picture of security, though. Understanding the whole picture of a corporate organization, and how to build an organization that responds to it. We will learn from Japanese cases and international standards.

Speakers: Shigeki Takei

Report

  • Next April, the company name will change to SCSK Net One Security
  • The vision is set as Rearchitecting Security
  • This is considered necessary for security in the era of frontier AI
  • A Threat Informed Approach is necessary
  • Why do attackers target my company?
    • There are indiscriminate attacks and targeted attacks, but perhaps it is because you matched their conditions
    • Rather than being targeted, it is a Condition match
    • Targets are shifting toward stopping the business rather than stopping the system
    • Because the return outweighs the cost spent on the attack
    • AI has not changed the attack methods
    • What attackers can do has increased
    • The cost of targeting has decreased
    • Ultimately, at the moment the attacker intrudes it is detected by the SOC/CSIRT, but in the reconnaissance phase before that, costs have been decreasing through the use of AI
  • Threat intelligence is for proactive defense
    • How things look from the attacker's perspective
    • What needs to be protected
    • We need to use AI to accelerate that verification
  • SCS Evaluation System
    • It is expected that evaluations such as 3 and 4 stars will begin operation
  • Vulnerability management is an important point
    • Business owners are now required to be able to explain to management what the risks are
    • Executives need to understand the risks and approve them
    • The SCS Evaluation System works in the direction of governance
  • ITU-T X.1060
    • A framework for entities that respond to cyber risk
      • Revised in 2026
    • There is also a Japanese version
  • CSC/CSC
    • The position of creating services
    • Service categories were set at 9
    • There is a classification of 64 services
    • Document layout
      • Design the system with security by design
    • Documents for building an organization
    • It is necessary to consider system operation together as well
    • Cybersecurity guidelines also include hands-on content
  • For the sake of security too, developers should make good use of AI
    • Have AI properly read the security documents and run with them
    • Those who manage systems should also discard old things, rebuild using AI, and test extensively
    • Things that are bad stay bad no matter how long you keep them
  • Message to executives
    • You are the one who decides

Impressions

We need to properly do what we can do.

We have to throw away all the legacy systems left untouched!

Share this article