2026年5月25日の
セキュリティ情報

Subject: [Security Alert] Typosquatting Attack in Go Module (shopsprint/decimal)

Dear IT/Security Team,

We are sharing information regarding a supply chain attack targeting the Go ecosystem.

■ Overview
A malicious package named 'github.com/shopsprint/decimal' was found mimicking the legitimate 'github.com/shopspring/decimal' library. The package implements a stealthy backdoor that polls DNS TXT records to receive and execute arbitrary remote commands (RCE). The attacker maintained the package as harmless for approximately six years before weaponizing it in August 2023 to evade detection.

■ Scope
- Affected Package: github.com/shopsprint/decimal
- Impacted Environments: Any development or production environment that imported this specific typosquatted module.

■ Mitigation Steps
1. Audit go.mod and go.sum files across all projects to ensure 'shopsprint/decimal' is not listed as a dependency.
2. If found, immediately remove the malicious package and replace it with the official 'github.com/shopspring/decimal'.
3. If the package was executed, assume that environment secrets (API tokens, SSH keys) may have been compromised and initiate a rotation of all sensitive credentials.

■ Reference
- Socket Security Research Report

Priority: High
Deadline: Immediate

Subject: [Info] Acceleration of Vulnerability Discovery via Claude Mythos AI

Dear Team,

We are sharing information regarding the acceleration of vulnerability discovery using Anthropic's AI model, "Claude Mythos."

■ Overview
Through "Project Glasswing," Anthropic and its partners (including Cloudflare, Mozilla, Microsoft, and Oracle) have identified tens of thousands of vulnerabilities. The AI's ability to find flaws in critical paths and open-source projects significantly exceeds human testing speeds.

■ Scope
- Firefox (Mozilla)
- Cloudflare systems
- Palo Alto Networks products
- Microsoft and Oracle products
- Over 1,000 open-source projects scanned by Anthropic

■ Recommended Actions
1. Ensure all critical software and OS are updated to the latest versions.
2. Monitor vendor security advisories closely, as the pace of vulnerability discovery and patching is expected to increase.

■ Reference
- Anthropic Official Blog

Priority: Low (Handle specific CVEs as per standard procedure)
Deadline: Ongoing

Subject: [Security Alert] Beware of AI-Powered Phishing and Fraud

Dear employees,

We would like to alert you to a rise in sophisticated fraud attacks using AI to create highly convincing messages that trick users into authorizing payments.

Requested Actions:
1. If you receive an urgent request for payment or money transfer, even from a known contact, please verify the request through a different communication channel (e.g., a phone call).
2. Do not click on suspicious links or open unknown attachments. Report any suspicious activity to the IT security team immediately.

Deadline: Ongoing awareness required

件名: 【共有】Visa 2026年春季半年度脅威レポート：AIによる行動リスクの増大について

お疲れさまです。Visaが発表した最新の脅威レポートに関する情報共有です。

■ 概要
技術的な防護（Token窃取や列挙攻撃への対策）が進んだ結果、攻撃者がAIを用いたソーシャルエンジニアリングにシフトし、ユーザーに「正当な権限で不正な決済を承認させる」手法が急増しています。防御の焦点は「認証情報の漏洩検知」から「欺瞞的な行動パターンの検知」へ移行する必要があります。

■ 影響範囲
- 決済システムを利用する全ユーザーおよび管理組織

■ 対応手順
1. ユーザー行動分析（UBA）などの行動ベースの検知メカニズムの検討
2. AI生成コンテンツによるフィッシングへの耐性を高めるための社員教育の更新
3. 決済承認フローにおける多層的な確認プロセスの再評価

■ 参考情報
- Visa Spring 2026 Biannual Threats Report

対応優先度: 中
対応期限: 次回セキュリティレビュー時まで

Subject: [Intel] Visa Spring 2026 Biannual Threats Report: Rise of AI-Driven Behavioral Risk

Dear Security Team,

This is a summary of the latest threat report from Visa.

■ Overview
As technical defenses against card enumeration and token theft improve, threat actors are pivoting toward AI-driven social engineering. The goal has shifted from stealing credentials to manipulating legitimate users into authorizing fraudulent transactions. Defense strategies must now evolve from credential detection to behavioral deception detection.

■ Scope
- All users and organizations utilizing payment systems.

■ Recommended Actions
1. Evaluate the implementation of User Behavior Analytics (UBA) to detect anomalous transaction patterns.
2. Update security awareness training to include AI-generated phishing examples.
3. Review and strengthen multi-step verification for high-risk payment authorizations.

■ Reference
- Visa Spring 2026 Biannual Threats Report

Priority: Medium
Deadline: Next security review cycle

Subject: [Technical Share] CISA ICS Security Advisories (AV26–506) Response

Dear Team,

This is a notification regarding the ICS security advisories published by CISA between May 18 and 24, 2026.

■ Overview
Multiple vulnerabilities have been identified in various Industrial Control Systems (ICS) products. Please refer to the specific advisories for CVE details and CVSS scores.

■ Affected Scope
- ABB B&R Automation Runtime / Automation Studio / PCs
- ABB CoreSense HM / M10
- ABB Terra AC Wallbox (JP)
- Hitachi Energy GMS600
- Kieback & Peter DDC Building Controllers
- ScadaBR
- Siemens RUGGEDCOM APE1808
- ZKTeco CCTV Cameras

■ Action Plan
1. Audit current assets to identify if any of the listed products/versions are in use.
2. Review the official CISA ICS Advisories for specific patches or mitigation steps.
3. Schedule and apply necessary updates according to maintenance windows.

■ Reference
- CISA ICS Advisories

Priority: High
Deadline: Immediate review and planned implementation