# Record Japan Cyberattack Surge: AI-Assisted Scanning and API Exploitation Drives Mass Email-Platform Data Exposure (Arara, Ymir Link/Cuenote FC)

> Japan recorded 233 publicly disclosed security incidents in October 2026 (about 2.4x its prior monthly peak), with roughly 185 million personal records exposed or potentially exposed. JPCERT/CC reported attackers using AI to scan for known weaknesses and attack APIs, and breaches at email distribution providers Arara (about 81.3 million addresses) and Ymir Link's Cuenote FC (about 47.9 million records) cascaded to downstream customers. No threat actor is publicly attributed, and AI's contribution to the overall surge is not established.

- **Published:** 2026-11-14T07:02:05Z
- **Last reviewed:** 2026-11-14T07:02:05Z
- **Canonical:** https://intel.threadlinqs.com/threat/TL-2026-6048
- **Card image:** https://intel.threadlinqs.com/og/threat/TL-2026-6048.png
- **ID:** TL-2026-6048
- **Severity:** HIGH
- **Category:** DATA_BREACH
- **Status:** ACTIVE
- **Detections:** 9 (1 shown free below) · **IOCs:** 10 (1 shown free below)

## Description

Japan's October 2026 disclosures reached 233 incidents, about 2.4 times the previous monthly record, per Trend Micro data reported by TV Asahi. MM Research Institute estimated about 185 million personal records were exposed or potentially exposed, nearly four times the 2025 total, and 30 incidents each involved at least one million records. These are record counts rather than unique individuals and mix suspected exposure with confirmed compromise. Macnica counted 288 web-system intrusions with potential data theft through October 2026 (168 in October alone) versus 84 in all of 2025. TrendAI counted 82 ransomware disclosures January-September 2026 versus 79 a year earlier, so the surge is driven by unauthorized access to web systems rather than ransomware.

JPCERT/CC's October 21 technical investigation confirmed that attackers used AI to scan for known vulnerabilities and attack APIs. They analyzed Android application packages to identify backend API endpoints and test them for flaws, including server-side request forgery (SSRF). An attack server examined by JPCERT/CC contained Blutter and jadx, tools for Flutter and Android reverse engineering that can expose authentication mechanisms and API routes not visible on public websites. TrendAI reported adaptive retries after initial failures and files with artifacts consistent with AI-generated code. Socket reported PHP vulnerability exploitation that began before the October 21 public disclosure. Researchers have not established how much AI contributed to the overall surge. Kyodo reported that GMO Internet Group's investigation pointed to a North Korea-linked group, but there is no public evidence tying the broader wave to one state-sponsored campaign, and no actor is attributed for the incidents below.

Ymir Link (Cuenote FC): Ymirlink detected an anomaly in an operations-management system account on October 16, blocked the attacker IP, and patched the entry-point vulnerability the same day. It isolated the server holding the unauthorized program on October 17 and reported to the Tokyo Metropolitan Police and the Personal Information Protection Commission on October 19. The vendor describes only a 'program vulnerability' (a software library vulnerability per Money Forward's summary); no product, CVE or version is named. Attackers placed programs on a server, obtained support and quality-control staff credentials, reached the core database and moved across the internal admin network. They also stole session information of logged-in users to reach the admin console and used the legitimate CSV export function, which resembled normal admin activity. The third report (November 11) puts exposure at 47,864,147 recipient records across 49 systems (down from a 51,628,280 maximum after deduplication), plus 226,158 hashed passwords and 420,419 authentication records for a discontinued customer-operated service. Cuenote BASE had 190 records viewable. Cuenote SR-S, SMS, Auth, Survey, Push and the safety-confirmation service were not affected. Downstream victims include NTTPC Communications (60,929 records).

Arara (Arara Message, operated under Paycloud Holdings): unauthorized access was confirmed on October 28. Attackers sent up to 7.13 million fraudulent emails from news@arara.com between October 28 and 30, impersonating credit card companies with subjects asking recipients to confirm account use or billing. Arara suspended the entire service on October 31 and disclosed the incident on November 2. Third-report figures (November 6) put exposure at about 81.3 million email addresses (81,329,467 per ScanNetSecurity) potentially affecting about 380 corporate customers, with up to 97 servers involved. Arara cites problems in its system environment and a web-server vulnerability, and an earlier shared-environment breach exposed 3,386,187 addresses from 61 companies. Exposed data varies by customer and may include names, birth dates, phone numbers, addresses and membership details. Downstream disclosures include SEGA (about 2.2 million newsletter addresses) and the Cabinet Office (62,863 addresses, 61,751 unique).

No CVE IDs, malware families, hashes, IPs or malicious domains are published in the sources reviewed. The only observable indicators are the abused sender domain/address and the attacker tooling named by JPCERT/CC. Defenders should treat this as a campaign-pattern and third-party-provider-risk threat: inventory software components, replace unsupported applications, prioritize and set deadlines for remediation, assess contractor-operated systems and APIs, and expect phishing that abuses legitimate provider domains. Japan announced a 10 billion yen emergency cybersecurity package on November 6.

## MITRE ATT&CK

- T1595.002 Vulnerability Scanning
- T1588.002 Tool
- T1190 Exploit Public-Facing Application
- T1078 Valid Accounts
- T1539 Steal Web Session Cookie
- T1550.004 Web Session Cookie
- T1550.004 Web Session Cookie
- T1213 Data from Information Repositories
- T1566 Phishing

## Free detection (Splunk SPL, Microsoft KQL, Sigma)

**TL-2026-6048 SSRF Payload Against API Endpoints (Cuenote FC / Arara Pattern) - Exploitation** (`TL-2026-6048-SPL-001`)

Flags web/API requests whose parameters embed loopback, RFC1918, link-local/cloud-metadata or non-HTTP scheme URLs. JPCERT/CC (Oct 21 2026) confirmed AI-assisted attackers tested backend APIs, discovered via Android APK analysis, for SSRF (CWE-918) during the Japan email-delivery platform intrusions (Ymirlink Cuenote FC, Arara Message).

- **Languages:** Splunk SPL, Microsoft KQL, Sigma
- **ATT&CK:** T1190, T1595.002
- **Data sources:** web access logs (index=web, sourcetype=access_combined_wcookie), IIS W3C logs (W3CIISLog), webserver logsource
- **Severity:** high
- **Confidence:** high

### Splunk SPL

```spl
index=web sourcetype=access_combined_wcookie earliest=-24h
| eval q=urldecode(coalesce(uri_query,""))
| where match(q,"(?i)(https?|gopher|file|dict|ftp)://(127[.]|localhost|0[.]0[.]0[.]0|10[.]|192[.]168[.]|172[.](1[6-9]|2[0-9]|3[01])[.]|169[.]254[.])")
| stats count as hits min(_time) as first_seen max(_time) as last_seen values(uri_path) as api_paths values(status) as statuses by clientip useragent
| eval first_seen=strftime(first_seen,"%F %T"), last_seen=strftime(last_seen,"%F %T")
| sort - hits
```

### Microsoft KQL

```kql
W3CIISLog
| where TimeGenerated > ago(24h)
| extend q = url_decode(csUriQuery)
| where q matches regex @"(?i)(https?|gopher|file|dict|ftp)://(127\.|localhost|0\.0\.0\.0|10\.|192\.168\.|172\.(1[6-9]|2[0-9]|3[01])\.|169\.254\.)"
| summarize Hits=count(), FirstSeen=min(TimeGenerated), LastSeen=max(TimeGenerated), ApiPaths=make_set(csUriStem, 20), Statuses=make_set(scStatus, 10) by cIP, csUserAgent
| order by Hits desc
```

### Sigma

```yaml
title: TL-2026-6048 SSRF Payload Targeting Internal or Metadata Addresses in API Request
id: 6b1f0c52-3a4e-4d7b-9c1a-2f5e8d6a1b01
status: experimental
description: Detects SSRF probing of web/API endpoints where a request parameter carries an internal, loopback, link-local metadata or non-HTTP scheme URL. JPCERT/CC confirmed attackers tested APIs found via APK analysis for SSRF in the Japan October 2026 email-platform intrusions (Cuenote FC, Arara).
references:
    - https://www.ymir.co.jp/info/ymir-information20261111.html
    - https://scan.netsecurity.ne.jp/article/2026/11/12/56759.html
    - https://gbhackers.com/japan-cyberattack-ai-security-vulnerabilities/
author: Threadlinqs Team
date: 2026/11/14
tags:
    - attack.initial_access
    - attack.t1190
    - attack.reconnaissance
    - attack.t1595.002
logsource:
    category: webserver
detection:
    selection:
        cs-uri-query|contains:
            - '=http://127.'
            - '=http://localhost'
            - '=http://169.254.'
            - '=http://10.'
            - '=http://192.168.'
            - '=https://169.254.'
            - '=gopher://'
            - '=file://'
            - '=dict://'
            - '%3Dhttp%3A%2F%2F127.'
            - '%3Dhttp%3A%2F%2Flocalhost'
            - '%3Dhttp%3A%2F%2F169.254.'
            - '%3Dhttp%3A%2F%2F10.'
            - '%3Dhttp%3A%2F%2F192.168.'
            - '%3Dgopher%3A%2F%2F'
            - '%3Dfile%3A%2F%2F'
    condition: selection
falsepositives:
    - Legitimate webhook or URL-preview features that fetch internal URLs
    - Authorized internal vulnerability scanners and pentests
level: high

```

**Known false positives:**

- Legitimate webhook, URL-preview or import-from-URL features that reference internal hosts
- Authorized vulnerability scanners and penetration tests

**Tuning notes:** Allowlist known scanner source IPs and sanctioned internal fetch endpoints by uri_path. Add your WAF/proxy index if requests terminate there. Extend the scheme list with IPv6 loopback and decimal/hex-encoded IPs if your stack decodes them.

### 8 more detection rules (Splunk SPL, Microsoft KQL, Sigma), unlock with TL Creds

- TL-2026-6048 Automated API Route Enumeration Burst - Exploitation (locked: unlock with TL Creds)
- TL-2026-6048 Bulk Admin CSV Export Burst (Cuenote FC Pattern) - Post-Exploitation (locked: unlock with TL Creds)
- TL-2026-6048 Credit-Card Impersonation Phishing via Abused [indicator] Sender - Delivery (locked: unlock with TL Creds)
- TL-2026-6048 Blutter / jadx Attacker Tooling Execution - Exploitation Preparation (locked: unlock with TL Creds)
- TL-2026-6048 Web/App Runtime Spawning Shell - Unauthorized Program on Server (locked: unlock with TL Creds)
- TL-2026-6048 Scripted Client Pulling Admin CSV Export - Data Collection (locked: unlock with TL Creds)
- TL-2026-6048 Admin Session Cookie Reuse Across Source IPs - Post-Exploitation (locked: unlock with TL Creds)
- TL-2026-6048 Single Account Authenticating to Many Hosts - Internal Lateral Movement (locked: unlock with TL Creds)

Sign up free to unlock the detection rules: https://threadlinqs.com/?modal=signup

## Sample indicators (1 of 10)

Defanged, TLP:CLEAR.

- Domain: `arara[.]com`

Full set of 10 indicators unlocks with TL Creds: https://threadlinqs.com/tl-creds

## Sources

- [Japan Hit by Record Cyberattack Surge as Hackers Use AI to Exploit Security Vulnerabilities (GBHackers)](https://gbhackers.com/japan-cyberattack-ai-security-vulnerabilities/)
- [Socket: Japan cyberattack surge analysis](https://socket.dev/blog/japan-cyberattack-surge)
- [Ymirlink third report on Cuenote FC unauthorized access (primary notice)](https://www.ymir.co.jp/info/ymir-information20261111.html)
- [SmartScope: Cuenote FC unauthorized access, third report (Nov 11, 2026)](https://smartscope.blog/en/blog/cuenote-fc-unauthorized-access-third-report-november-11-2026/)
- [Money Forward Admina: Ymirlink Cuenote unauthorized access summary](https://admina.moneyforward.com/jp/blog/ymirlink-cuenote-unauthorized-access)
- [SmartScope: NTTPC Communications 60,929 customer email addresses leaked via Cuenote FC](https://smartscope.blog/en/blog/nttpc-communications-60-929-customer-email-addresses-leaked-2026/)
- [eSecurityPlanet: Arara Message Hack, 77 Million Emails Exposed, Millions of Phishing Messages Sent](https://www.esecurityplanet.com/threats/news-arara-message-breach-77-million-email-addresses-japan-apac/)
- [ScanNetSecurity: Arara Message unauthorized access, 81,329,467 email addresses leaked (third report)](https://scan.netsecurity.ne.jp/article/2026/11/12/56759.html)
- [SmartScope: Cabinet Office email delivery service Arara, 62,863 addresses leaked](https://smartscope.blog/blog/cabinet-office-email-delivery-service-arara-62-863-2026/)

## Full data

The other 8 detection rules (Splunk SPL / Microsoft KQL / Sigma) unlock with TL Creds; a free account includes daily creds: https://threadlinqs.com/?modal=signup
Indicator values unlock with TL Creds, at a lower price on Red and Purple; AI agents can query rules and IOCs through the Threadlinqs MCP server (Purple): https://intel.threadlinqs.com/mcp

Canonical: https://intel.threadlinqs.com/threat/TL-2026-6048
