TI-2026-019D โ "The Open Doors" โ 156 MySQL, 74 PostgreSQL, 5 MongoDB Databases Exposed on Compromised Infrastructure Attacking the Internet
๐ The Dual Threat
This dossier presents the other side of the coin. Parts A-C documented who is attacking and how they profit. Part D asks: what are the attackers leaving behind, and what data is exposed on the infrastructure they've already compromised?
Among the 7,614 IPs actively attacking our honeypot, Shodan reveals that 156 expose MySQL, 74 expose PostgreSQL, 5 expose MongoDB, 19 expose RDP, and 13 expose Redis โ all on IPs with threat scores above 50. These are not just attackers. These are compromised systems with databases that are simultaneously attacking others AND exposing their own data.
When those databases belong to universities (student records), telecoms (subscriber data), hospitals (medical records), or government agencies (citizen data) โ the human cost of a single SSH brute-force becomes catastrophic.
๐ซ SEGMENT 1: Education Networks โ Student Data at Risk
| Institution | Country | Threat | Exposed Services | Impact |
|---|---|---|---|---|
| Hanoi University of Science & Technology | ๐ป๐ณ | 100 | 443 (hosts ACM AsiaCCS 2025 cybersecurity conference!) | A cybersecurity conference website attacking the internet |
| TERNET (Tanzania, 67+ universities) | ๐น๐ฟ | 100 | SSH only (41.93.28.9) | Plants mdrfckr backdoor globally |
| ZAMREN (Zambia Research/Education) | ๐ฟ๐ฒ | 100 | 80, 2222, 8000 | Web + alt-SSH + dev server exposed |
| Koforidua Technical University | ๐ฌ๐ญ | 100 | 21, 161, 443, 2000, 8291, 8728 | MikroTik router directly compromised |
| RENATA (Colombia Academic Network) | ๐จ๐ด | 100 | 22, 25, 443, 8080 | Email + web + proxy exposed |
| Kajeh Nasir Toosi University | ๐ฎ๐ท | 50 | โ | Iranian university attacking |
Hanoi University of Science and Technology (HUST) โ IP 202.191.58.34 โ is simultaneously:
- Hosting asiaccs2025.hust.edu.vn โ the ACM Asia Conference on Computer and Communications Security 2025 website
- Hosting bkcs.hust.edu.vn โ the university's Computer Science department
- Registered as a threat score 100 attacker in our honeypot
A cybersecurity conference's own hosting infrastructure is compromised and attacking the internet. The people presenting papers about defending against SSH attacks are hosted on infrastructure that IS an SSH attack.
When a university network is compromised by botnet malware (confirmed: TERNET, HUST, ZAMREN, Koforidua, RENATA), the attacker has access to:
- Student Information Systems โ names, national IDs, addresses, phone numbers, grades, financial aid status
- Research data โ unpublished papers, grant proposals, experimental data, intellectual property
- Email systems โ all communications between faculty, students, and administration
- Medical records โ university health centers (TERNET connects medical universities)
- Financial systems โ tuition payments, salary information, bank account details
For Koforidua Technical University (Ghana): the exposed MikroTik API port (8728) means the router itself is compromised. This enables: traffic interception of ALL university traffic, DNS poisoning to redirect students to phishing sites, and credential capture at the network layer.
Total students potentially affected: TERNET alone connects 67+ institutions with estimated 200,000+ students. HUST has 40,000+ students. ZAMREN serves all Zambian universities. Conservative estimate: 500,000+ student records accessible through compromised education networks in our dataset.
๐ฅ SEGMENT 2: Viettel Military โ Databases on a Military Network
Three Viettel Group IPs expose database ports to the public internet:
| IP | Threat | Exposed Ports | Risk |
|---|---|---|---|
171.244.37.97 | 50 | 22, 443, 5005, 27017 | MongoDB (often no auth by default) |
171.244.37.103 | 50 | 22, 5432 | PostgreSQL exposed |
125.212.243.14 | 44 | 21, 53, 80, 443, 3306, 4444 | MySQL + Metasploit (4444)! |
IP 125.212.243.14 (Viettel, Vietnam People's Army) exposes port 4444 โ the default Metasploit reverse shell listener port. Combined with MySQL 3306 and FTP 21, this is either:
- A compromised server running Metasploit payloads โ the botnet operator has planted a persistent reverse shell
- An offensive security tool deliberately deployed โ Viettel's Cyber Security division (VCS) conducting operations
Either answer is alarming. If compromised: a foreign criminal group has a Metasploit shell on Vietnam's military telecom with database access. If deliberate: Vietnam's military is running offensive tools that are simultaneously attacking our honeypot.
What Viettel databases might contain:
Viettel serves 77 million subscribers in Vietnam alone, plus operations in 10 countries (Myanmar, Laos, Cambodia, Haiti, Mozambique, Tanzania, Burundi, Cameroon, Peru, East Timor). A compromised database on Viettel infrastructure could expose:
- Subscriber identity data (required by Vietnamese law for SIM registration)
- Call detail records (CDRs) โ who called whom, when, for how long
- Location data from cell towers
- Mobile payment/Viettel Pay transaction records
- In countries like Myanmar: opposition movement communications during military coup
๐ SEGMENT 3: Korea Telecom โ A Nation's Database Ports Wide Open
Korea Telecom (AS4766) โ serving 90% of South Korea's fixed-line internet โ has 90 IPs in our dataset with an extraordinary range of exposed services:
Critical Exposed Ports on Korea Telecom IPs (attacking our honeypot):
1521 โ Oracle Database3306 โ MySQL5432 โ PostgreSQL3389 โ RDP (Remote Desktop)1883 โ MQTT (IoT)5900 โ VNC (screen share)6379 โ Redis (cache)8333 โ Bitcoin node7547 โ TR-069 (ISP remote management)Korea Telecom's exposed services tell a story of massively compromised residential and small-business devices:
- Port 7547 (TR-069) โ ISP remote management protocol. If exposed and compromised, an attacker can remotely reconfigure every customer's router connected to this management server. The Mirai botnet exploited exactly this vulnerability in the 2016 Deutsche Telekom attack (900,000 routers offline).
- Port 8333 (Bitcoin) โ A Korea Telecom customer running a Bitcoin full node. If this device is compromised and attacking our honeypot, the Bitcoin wallet on it is also compromised.
- Port 1883 (MQTT) โ IoT messaging protocol. Smart home devices, industrial sensors, medical devices. All connected, all potentially leaking data through a compromised gateway.
- Oracle Database (1521) on a Korea Telecom IP at threat score 100 โ this is likely a business server (small/medium enterprise) with customer data, financial records, or inventory systems. Compromised AND attacking the internet simultaneously.
South Korea context: With 52 million population and 90% KT market share, approximately 47 million people depend on Korea Telecom infrastructure. The 90 compromised IPs we see are the tip โ they're the ones actively attacking. The compromised devices NOT attacking (just mining, or dormant) are invisible to us.
๐ SEGMENT 4: The Cascade Effect โ When Attackers ARE the Vulnerable
The traditional security model assumes attackers and victims are separate. Our data reveals they're often the same machine. A server compromised by the Outlaw botnet becomes BOTH:
- An attack platform (scanning, brute-forcing, planting backdoors)
- A data exposure point (its own databases, services, and user data are accessible to the botnet operator AND to anyone else who scans it)
| Database servers exposed in our dataset (threat > 50) | 250+ |
| Average records per small business database | 10,000 โ 100,000 |
| Average records per telecom subscriber DB | 1,000,000+ |
| Average records per university SIS | 20,000 โ 200,000 |
| Conservative total data exposure estimate | 2.5M โ 25M records |
This is from OUR dataset of 7,614 IPs alone. We observe a single honeypot. The actual number of compromised databases globally is orders of magnitude larger.
๐ถ SEGMENT 5: The Children's Data Connection
When education networks serving children and young adults are compromised, the exposed data has unique vulnerability characteristics that distinguish it from typical breaches:
- Long exploitation window: A stolen adult identity averages 2 years before detection. A stolen child's identity averages 7-10 years โ because children don't check credit reports, don't apply for loans, and parents rarely monitor child credit.
- Clean credit history: Children's SSN/national IDs have zero credit history โ perfect for creating synthetic identities. Each child's record is worth $25-50 on dark web markets vs. $1-5 for adults.
- Permanent damage: By the time a child turns 18 and applies for their first loan or rental, their credit may already be destroyed. Remediation takes 2-5 years of disputes.
- COPPA/GDPR-K violations: Organizations storing children's data under COPPA (US) or GDPR Article 8 (EU) face enhanced fines for breaches โ up to $50,000 per child in the US. But when the breach is in Tanzania, Ghana, or Vietnam โ there IS no enforcement.
TERNET (Tanzania): Connects universities but ALSO teacher training colleges and secondary school networks. Estimated 200,000+ students, many under 18. Their data is accessible to whoever controls the mdrfckr backdoor on 41.93.28.9.
Koforidua Technical University (Ghana): Router-level compromise (MikroTik API exposed) means ALL traffic โ including student registration, exam results, financial aid applications โ can be intercepted in real-time.
- Data protection laws exist on paper but lack enforcement capacity
- Security budgets are below subsistence level (TERNET CEO: "nearly failed to pay staff")
- International development funding built the networks but didn't fund ongoing security
- The Budapest Convention doesn't apply
- Children and young adults have the LEAST ability to detect or respond to identity theft
โ๏ธ SEGMENT 6: Who Is Responsible?
The chain of responsibility for a compromised education network:
| Actor | Responsibility | Accountability |
|---|---|---|
| The Attacker (Outlaw gang) | Committed the intrusion | Zero โ operating from Romania/Portugal, no extradition treaties with victims |
| The Institution (e.g., TERNET) | Failed to secure infrastructure | Minimal โ no budget, no staff, no regulatory enforcement |
| The Donor (EU AfricaConnect2) | Built infrastructure without security | None โ project delivered, grant closed, no ongoing obligation |
| The Transit Provider | Carries attack traffic | None โ common carrier protections |
| The Cloud Provider (hosting C2) | Hosts command infrastructure | None โ shared responsibility model |
| The Regulator | Should enforce data protection | Non-existent in most affected jurisdictions |
| The Students/Children | None whatsoever | Bear 100% of the consequences |
The accountability gap is total. Every actor in the chain has either legal immunity, jurisdictional protection, or insufficient resources to be held responsible. The only party that bears actual consequences is the one with zero agency in the matter: the students whose data is exposed.
This is not a technical problem. It's a systemic design choice. The internet was built without accountability architecture. Every layer optimizes for throughput, not safety. And the cost is paid by those least able to afford it.