๐ TI-2026-023D โ The Trust Chain: How 1,357 Geographic Lies Cascade Through Every Database on Earth
Executive Summary
When five different databases are asked "Where is this IP address?" โ they give five different answers. Not occasionally. Not for edge cases. For 1,357 IP addresses in our threat intelligence database โ 17.1% of all tracked IPs โ at least two authoritative sources disagree about which country they belong to.
This is not a measurement error. It is a structural failure in how the internet knows where things are. BGP routing says one country. RDAP registration says another. AbuseIPDB agrees with a third. Shodan has a fourth opinion. IPinfo offers a fifth โ and flags zero of these discrepant IPs as VPN, proxy, or suspicious.
Every compliance system, every sanctions screen, every content restriction, every law enforcement investigation, every advertising geo-target, and every court warrant that depends on "knowing" where an IP is โ all of them consume these databases. When the databases are wrong, everything downstream breaks. Wrong doors get kicked in. Sanctions get evaded. Laws get broken in countries that don't know it's happening. This dossier traces the trust chain from source to consequence.
Chapter 1: The Five Opinions Problem
Ask any security researcher, compliance officer, or law enforcement agent where an IP address is located, and they'll consult a GeoIP database. They believe they're getting a fact. They're getting an opinion โ and depending on which database they ask, they may get a different opinion every time.
Our honeypot threat intelligence platform enriches every attacker IP from multiple independent sources. Each source uses a different methodology to determine geographic location:
- BGP/Cymru โ Where the IP is routed (which country's network infrastructure carries the traffic)
- RDAP/WHOIS โ Where the IP is registered (the legal entity that holds the allocation from a Regional Internet Registry)
- AbuseIPDB โ Where the IP is reported (community consensus based on MaxMind lookups)
- Shodan โ Where the IP appears to be (based on scanning, banners, and service fingerprints)
- IPinfo โ Where the IP is commercially determined (proprietary algorithm combining BGP, DNS, traceroute, and carrier data)
Each source has legitimate reasons for its answer. BGP reflects physical routing. RDAP reflects legal ownership. AbuseIPDB reflects the consensus of millions of reports. Shodan reflects what's actually running. IPinfo reflects commercial intelligence.
The problem is not that one source is wrong. The problem is that they're all measuring different things โ and every downstream consumer treats them as if they're measuring the same thing.
๐ด CRITICAL FINDING: Source Agreement Failure
Across 2,214 IPs where we have data from all three primary sources (AbuseIPDB, BGP/Cymru, and RDAP):
| Source | Agrees With Derived Country | Rate | Methodology |
|---|---|---|---|
| AbuseIPDB | 2,214 / 2,214 | 100.0% | MaxMind-based geolocation |
| RDAP | 1,901 / 2,214 | 85.9% | Legal registration |
| BGP/Cymru | 1,518 / 2,214 | 68.6% | Routing infrastructure |
AbuseIPDB agrees 100% because it is the primary input to the derived country field. This is not independent confirmation โ it is circular validation. The same MaxMind database that feeds AbuseIPDB feeds the compliance systems that trust it. When MaxMind is wrong, the entire chain is wrong, and nobody notices because every link reports the same incorrect answer.
Chapter 2: The Geography Corridors โ Where Countries Move
Discrepancies are not random. They cluster into corridors โ repeated patterns where the same type of geographic mismatch appears across dozens of IPs. These corridors tell a story about how infrastructure is actually organized versus how it is presented.
| BGP Says | RDAP Says | IPs | Pattern | |
|---|---|---|---|---|
| ๐ธ๐ฌ Singapore | โ | ๐บ๐ธ United States | 41 | Tencent Cloud virtual US instances from SG infrastructure |
| ๐ฆ๐ช UAE | โ | ๐ฎ๐ถ Iraq | 39 | Iraqi ISP routing through UAE transit โ conflict zone pattern |
| ๐ฌ๐ง UK | โ | ๐บ๐ธ United States | 33 | Cloud providers with UK BGP, US legal entities |
| ๐ฉ๐ช Germany | โ | ๐บ๐ธ United States | 23 | US companies hosting in German data centers |
| ๐ณ๐ฑ Netherlands | โ | ๐ซ๐ท France | 23 | French companies using Dutch infrastructure (OVH pattern) |
| ๐ฉ๐ช Germany | โ | ๐ซ๐ฎ Finland | 19 | Hetzner Finland marketed as German hosting |
| ๐ธ๐ฌ Singapore | โ | ๐ญ๐ฐ Hong Kong | 13 | Tencent/Yisu Cloud SG-HK routing |
| ๐ฌ๐ง UK | โ | ๐ฉ๐ช Germany | 13 | German entities routing through London |
| ๐ธ๐จ Seychelles | โ | ๐ญ๐ฐ Hong Kong | 13 | Yisu Cloud / Cloud Innovation Support โ offshore shell |
| ๐ธ๐ฌ Singapore | โ | ๐ฎ๐ฉ Indonesia | 12 | Tencent Cloud Indonesia via SG transit |
| ๐ท๐บ Russia | โ | ๐ต๐ฑ Poland | 8 | Russian operations through Polish RDAP registration |
| ๐บ๐ฆ Ukraine | โ | ๐ณ๐ฑ Netherlands | 7 | Ukrainian operators using Dutch registrations |
Reading the Corridors
Each corridor has a different meaning:
๐ธ๐ฌโ๐บ๐ธ The Cloud Arbitrage Corridor (41 IPs)
Tencent Cloud (AS132203, operating as "AceVille Pte Ltd" in Singapore) sells "US servers" that are physically routed through Singapore. BGP knows the traffic traverses SG networks. RDAP says the allocation belongs to a US entity. GeoIP databases can't agree because the server is simultaneously in both countries โ physically in Singapore's network fabric, legally in America's registry, and marketed as a US endpoint to customers.
This is the cloud industry's fundamental geographic fiction: a server's "location" is wherever you want it to be.
๐ฆ๐ชโ๐ฎ๐ถ The Conflict Zone Corridor (39 IPs)
All 39 IPs belong to a single entity: Hulum Almustakbal Company for Communication Engineering and Services Ltd (AS203214), an Iraqi telecommunications company. Their traffic routes through UAE โ presumably because Iraq's domestic internet infrastructure, damaged by decades of conflict, relies on UAE as a transit hub.
GeoIP databases see UAE routing and assign UAE as the location. RDAP correctly identifies Iraqi registration. The result: every GeoIP-based compliance screen treats Iraqi internet traffic as if it comes from the Emirates. For sanctions enforcement, for counter-terrorism monitoring, for any system that cares about the distinction between Iraq and UAE โ the databases are systematically wrong.
๐ธ๐จโ๐ญ๐ฐ The Offshore Laundering Corridor (13 IPs)
BGP says Seychelles. RDAP says Hong Kong. The organization: "Cloud Innovation Support" โ a front for Yisu Cloud, a hosting provider we've encountered across multiple dossiers. The SeychellesโHong Kong corridor is the geographic signature of offshore shell company infrastructure: register in the Seychelles for anonymity, route through Hong Kong for connectivity, appear in GeoIP databases as whichever story is more convenient.
๐ Conspirative Question: The Perfect Laundering Route
If you wanted to operate internet infrastructure that couldn't be attributed to any single jurisdiction, you would:
- Register through an offshore jurisdiction (Seychelles, BVI, Belize) โ no transparency requirements
- Route BGP through a different country (Singapore, UAE, Netherlands) โ different legal jurisdiction for traffic interception
- File RDAP in a third country (Hong Kong, US, Germany) โ different jurisdiction for registration inquiries
- Publish geofeed claiming a fourth country โ GeoIP databases report yet another location
The result: four different countries involved, no single authority with complete visibility, and every investigator sees a different piece of the puzzle depending on which database they query first.
Chapter 3: The Organizations That Can't Be Pinned Down
Geographic discrepancies concentrate in specific organizations. The top 10 organizations account for over 300 discrepant IPs โ and each one tells a story about why geography breaks.
| Organization | Discrepant IPs | Pattern | Risk Assessment |
|---|---|---|---|
| RIPE-NCC-HM-MNT | 94 | Default RIPE maintainer โ no specific org claimed | HIGH: Anonymous registrations |
| AceVille Pte Ltd (Tencent) | 53 | SG BGP, multi-country RDAP | HIGH: Cloud geographic fiction |
| Contabo GmbH | 47 | German hosting, international routing | MEDIUM: Budget hosting provider |
| OVH / OVHcloud | 20 | French company, multi-country infrastructure | MEDIUM: Legitimate cloud provider |
| Cloud Innovation Support | 19 | Yisu Cloud front โ SC/HK/CN | CRITICAL: Offshore shell |
| "Private Customer" | 17 | No org disclosed in RDAP | HIGH: Deliberately anonymous |
| serversf-mnt (Servers Factory) | 17 | GB BGP, US RDAP | HIGH: Known attack infrastructure |
| Hetzner Online GmbH | 16 | DE company with FI infrastructure | MEDIUM: Legitimate provider |
| ZWIEBELFREUNDE | 16 | Tor-associated, multi-country | HIGH: Privacy infrastructure |
| AS8560-MNT (IONOS/1&1) | 15 | German parent, international hosting | MEDIUM: Large provider |
The Anonymous Majority
The #1 organization with geographic discrepancies is RIPE-NCC-HM-MNT โ which is not an organization at all. It's the default maintainer RIPE assigns when no specific entity is claimed. Ninety-four IPs with country discrepancies belong to registrations that don't name their operator.
Combined with "Private Customer" (17 IPs) and various other non-descriptive maintainers, over 130 discrepant IPs โ nearly 10% of all discrepancies โ belong to entities that refuse to identify themselves.
๐ HIGH FINDING: Anonymity Correlates With Discrepancy
IPs registered to anonymous or minimal RDAP entities are significantly more likely to have geographic discrepancies. This is not coincidental โ the same motivation that drives anonymous registration (avoiding attribution) drives geographic obfuscation. Both serve the same purpose: making infrastructure impossible to investigate.
Chapter 4: The Tencent Cloud Case Study โ 76 IPs, 12 Countries, One Singapore Rack
AS132203 operates under the name "AceVille Pte Ltd" โ a Singapore entity that is, in reality, Tencent Cloud International, the overseas arm of China's internet giant Tencent Holdings (HKEX: 0700, market cap ~$450B).
Tencent Cloud sells virtual servers in 12+ countries. All BGP routing for these servers goes through Singapore. The geographic distribution of our 76 tracked AS132203 IPs:
| Derived Country | BGP | RDAP | IPs | Avg Threat | Hits |
|---|---|---|---|---|---|
| ๐ธ๐ฌ Singapore | SG | SG | 71 | 57.7 | 228 |
| ๐บ๐ธ United States | SG | US | 32 | 58.0 | 97 |
| ๐บ๐ธ United States | CN | US | 8 | 54.0 | 0 |
| ๐ฏ๐ต Japan | SG | JP | 6 | 62.0 | 107 |
| ๐ญ๐ฐ Hong Kong | SG | HK | 6 | 54.7 | 5 |
| ๐ฉ๐ช Germany | SG | DE | 3 | 64.3 | 39 |
| ๐ง๐ท Brazil | SG | BR | 3 | 54.7 | 4 |
| ๐ฎ๐ฉ Indonesia | SG | ID | 2 | 48.0 | 0 |
| ๐ฎ๐ณ India | SG | IN | 2 | 75.0 | 77 |
| ๐ฐ๐ท South Korea | SG | KR | 1 | 85.0 | 21 |
| ๐น๐ญ Thailand | SG | TH | 2 | 48.0 | 0 |
The "US" servers in Tencent Cloud are interesting. BGP says Singapore. RDAP says United States. IPinfo agrees: United States. The customer buying a "US server" gets a Singapore-routed endpoint with a US IP address โ the GeoIP database says "US," every website and compliance system treats it as US traffic, but the actual packets traverse Singapore.
๐ The Cascade: What Breaks When Tencent's "US" Servers Are in Singapore
- Data residency: EU companies using Tencent's "US" servers for GDPR-compliant processing are actually routing data through Singapore โ subject to Singapore's Cybersecurity Act and potential access by Chinese authorities (Singapore's PDPA has different protections than US frameworks)
- Content licensing: Netflix, Hulu, and other streaming services geo-restrict by GeoIP. A "US" Tencent IP gets US content โ even though the physical infrastructure is in Asia
- Financial compliance: Trading platforms that restrict by jurisdiction treat Tencent "US" IPs as US-based. Users can appear to comply with US trading hours and regulations while actually operating from Asia
- Law enforcement: A warrant for "the server at this US IP address" goes to a US-registered entity โ but the physical hardware is in Singapore, requiring mutual legal assistance treaties (MLATs) that may take months or years
๐ Reading Between the Lines
Tencent doesn't call itself Tencent in Singapore. It operates as "AceVille Pte Ltd" โ a generic, unmemorable company name that doesn't trigger the same regulatory attention as "Tencent Cloud, subsidiary of China's largest internet conglomerate." The name change is itself a form of geographic obfuscation: not physical, but institutional. A compliance officer seeing "AceVille Pte Ltd" in RDAP doesn't know they're dealing with a Tencent subsidiary. They have to trace the corporate chain โ and most don't.
Why does this matter? Because Tencent Holdings is subject to US Executive Order 14032 (Chinese Military-Industrial Complex Companies Securities Investments) restrictions. A company under US investment sanctions operating cloud infrastructure that GeoIP databases classify as "US-based" creates a compliance paradox that few systems are designed to detect.
Chapter 5: The Iraq-UAE Corridor โ A Case Study in Conflict Zone Geography
Thirty-nine IPs share the same pattern: BGP says UAE, RDAP says Iraq. All belong to Hulum Almustakbal Company for Communication Engineering and Services Ltd (AS203214) โ "Hulum" meaning "dream" in Arabic. Their average threat score: 63.8.
This is what happens when a country's internet infrastructure is destroyed by war and rebuilt through neighboring states. Iraq's internet has been rebuilt largely through transit agreements with Turkey, Kuwait, and the UAE. Hulum Almustakbal routes through UAE infrastructure because it's faster, cheaper, and more reliable than Iraqi domestic fiber.
But GeoIP databases don't understand geopolitics. They see UAE routing and assign UAE. The result:
๐ HIGH FINDING: Conflict Zone Geographic Misattribution
39 Iraqi IPs are systematically classified as UAE by GeoIP databases that rely on BGP routing. For any system that cares about the distinction โ counter-terrorism screening, sanctions compliance, intelligence analysis โ Iraqi internet traffic is invisible. It appears as UAE traffic, and no flag is raised.
This is not unique to Iraq. Any country that relies on neighboring transit infrastructure โ Syria through Turkey, Yemen through Saudi Arabia, Afghanistan through Pakistan โ suffers the same misattribution. The countries most important to monitor are the ones most likely to be invisible in GeoIP databases.
๐ Conspirative Question: Who Benefits From Iraqi Traffic Looking Emirati?
If you were conducting surveillance on Iraqi internet traffic, you would need cooperation from Iraqi authorities โ or you would need the traffic to transit through a country whose intelligence services you do cooperate with. UAE is a member of the Fourteen Eyes intelligence alliance (via bilateral agreements). Iraq is not.
The routing of Iraqi traffic through UAE is explained by infrastructure economics. But it is also convenient for signals intelligence: 39 Iraqi IPs worth of traffic traversing UAE networks, where it can be examined under UAE's Federal Decree-Law No. 34/2021 on Combating Rumours and Cybercrime, which allows broad interception authority.
Chapter 6: The Invisible Discrepancy โ Why IPinfo Sees Nothing
IPinfo is one of the most respected GeoIP providers. It enriches IP addresses with geolocation, ASN information, and critically โ privacy detection flags: VPN, proxy, Tor, relay, and hosting.
We have IPinfo enrichment for 600 of our discrepant IPs โ IPs where at least two authoritative sources disagree about the country. The results:
| IPinfo Flag | Count | Percentage |
|---|---|---|
| VPN detected | 0 | 0.0% |
| Proxy detected | 0 | 0.0% |
| Hosting detected | 0 | 0.0% |
| Tor detected | 0 | 0.0% |
Zero. Not a single discrepant IP is flagged by IPinfo as VPN, proxy, hosting, or any type of privacy service. IPinfo's privacy detection โ which compliance systems worldwide rely on โ is completely blind to geographic discrepancies.
๐ด CRITICAL FINDING: Privacy Detection Blind Spot
IPinfo's commercial privacy detection product identifies zero (0) of 600 geographically discrepant IPs as VPN, proxy, or hosting infrastructure. This means:
- A compliance system using IPinfo to detect proxy/VPN usage will not flag traffic from IPs where multiple sources disagree about country
- Anti-fraud systems using IPinfo will trust the stated location of IPs that are demonstrably mislocated
- Financial KYC checks will accept geographic claims from IPs where the physical routing contradicts the GeoIP assignment
- The companies paying IPinfo for "privacy detection" are paying for a false sense of security
Source: Our enrichment database, cross-referencing deep_ipinfo VPN/proxy/hosting/tor flags against country discrepancy analysis. Sample: 600 IPs with IPinfo data and BGPโ RDAP disagreement.
๐ Reading Between the Lines
IPinfo's VPN detection works by identifying known VPN provider IP ranges. If NordVPN uses M247's IP 95.181.237.131 as a "Venezuela" endpoint, IPinfo may or may not flag it as VPN โ but it certainly doesn't flag it as "discrepant geography." The VPN detection and the geographic data are separate products that don't talk to each other.
This is by design, not by accident. If IPinfo flagged every IP with a geographic discrepancy as suspicious, they'd flag 17% of our database โ including legitimate cloud providers like Tencent, OVH, Hetzner, and Microsoft. Their customers would complain. So the discrepancy goes unreported, and the trust chain remains intact โ built on silence rather than accuracy.
Chapter 7: The ASN Discrepancy Ranking โ Which Networks Lie Most
Some networks produce more geographic discrepancies than others. The top 15 ASNs by discrepancy count:
| ASN | Operator | Discrepant IPs | Pattern |
|---|---|---|---|
| AS132203 | Tencent Cloud / AceVille Pte Ltd | 76 | Multi-country cloud from SG |
| AS203214 | Hulum Almustakbal (Iraq) | 37 | Iraq via UAE transit |
| AS9009 | M247 Europe SRL | 30 | Virtual location manufacturing |
| AS209334 | Servers.com (NL) | 23 | Multi-country hosting |
| AS16276 | OVHcloud | 22 | French company, global infra |
| AS8560 | IONOS / 1&1 | 19 | German company, multi-country |
| AS14061 | DigitalOcean | 18 | US company, global datacenters |
| AS24940 | Hetzner Online | 18 | DE/FI infrastructure |
| AS215540 | Unknown operator | 17 | Unidentified โ newly allocated ASN |
| AS60729 | Zwiebelfreunde e.V. | 16 | Tor-associated nonprofit |
| AS135377 | UCLOUD HK Limited | 13 | Chinese cloud via HK |
| AS25369 | Hydra Communications | 13 | UK-based VPN/hosting |
| AS210558 | Unknown operator | 12 | Recently allocated ASN |
| AS8075 | Microsoft | 12 | Azure global infrastructure |
| AS150436 | Unknown operator | 11 | Asian cloud provider |
Three categories emerge:
- Legitimate cloud providers (Tencent, OVH, Hetzner, DigitalOcean, Microsoft) โ discrepancies caused by global infrastructure that GeoIP can't accurately classify
- Infrastructure conflict zones (Hulum/Iraq, UCLOUD/HK) โ discrepancies caused by geopolitical reality that GeoIP doesn't model
- Deliberate obfuscation (M247, Zwiebelfreunde, unknown ASNs) โ discrepancies that are the product, not the side effect
The critical insight: GeoIP databases treat all three categories identically. They can't distinguish between Microsoft's Azure having a server in a UK datacenter (benign) and M247 manufacturing a "virtual Iran" from Romania (potentially illegal). Both appear as a country discrepancy. Both are reported with the same confidence.
Chapter 8: The Multi-Source Matrix โ When Every Database Disagrees
For the most dangerous discrepant IPs (threat score โฅ 85), we can compare five sources simultaneously. The result is a confusion matrix where no two sources fully agree:
| IP | Score | Derived | AbuseIPDB | BGP | RDAP | IPinfo | Org |
|---|---|---|---|---|---|---|---|
| 185.222.138.237 | 97 | ๐ฝ๐ฐ XK | ๐ฝ๐ฐ XK | ๐ฉ๐ช DE | ๐ฆ๐ฑ AL | ๐ฝ๐ฐ XK | TERRATRANSIT-MNT |
| 87.251.64.176 | 99 | ๐ต๐ฑ PL | ๐ต๐ฑ PL | ๐ท๐บ RU | ๐ต๐ฑ PL | ๐ต๐ฑ PL | ru-avm-1-mnt |
| 154.209.4.195 | 98 | ๐ญ๐ฐ HK | ๐ญ๐ฐ HK | ๐ธ๐จ SC | ๐ญ๐ฐ HK | ๐ญ๐ฐ HK | Cloud Innovation Support |
| 179.61.232.244 | 98 | ๐บ๐ธ US | ๐บ๐ธ US | ๐ฆ๐ช AE | ๐บ๐ธ US | ๐บ๐ธ US | "Private Customer" |
| 185.246.128.133 | 95 | ๐ธ๐ช SE | ๐ธ๐ช SE | ๐ฌ๐ง GB | ๐ธ๐ช SE | ๐ธ๐ช SE | lir-uk-win-1-MNT |
| 192.250.235.126 | 95 | ๐ธ๐ฌ SG | ๐ธ๐ฌ SG | ๐ฌ๐ง GB | ๐ธ๐ฌ SG | ๐ธ๐ฌ SG | WHG Hosting Services |
The Kosovo Paradox
IP 185.222.138.237 (threat score 97) is classified as Kosovo by three sources โ but Kosovo is not universally recognized as a country. Serbia, Russia, China, and 87 other UN member states don't recognize it. RDAP registers it to Albania. BGP routes through Germany. The GeoIP classification as "Kosovo" (XK) is itself a political statement that databases make without acknowledging.
The TERRATRANSIT-MNT maintainer operates in the Kosovo-Albania-Germany triangle โ the geographic discrepancy reflects not just technical infrastructure but unresolved geopolitical status encoded into IP geolocation.
The "Private Customer" Pattern
IP 179.61.232.244 (threat score 98) belongs to a "Private Customer" โ RDAP reveals nothing about the operator. It's registered to the US, routes through UAE, and has one of the highest threat scores in our database. This is the convergence of anonymity and geographic obfuscation: you can't identify who runs it AND you can't determine where it is.
๐ Conspirative Question: The Russian Maintainer in Poland
IP 87.251.64.176 (threat score 99) is registered in Poland, routes through Russia, and its RDAP maintainer is named "ru-avm-1-mnt" โ the "ru-" prefix explicitly referencing Russia. Every GeoIP database says Poland. BGP says Russia. The maintainer's own name says Russia.
After February 2022, RIPE announced measures regarding Russian entities, but allowed existing registrations to continue. An operator named "ru-avm" maintaining Polish-registered infrastructure that routes through Russia is exactly the kind of sanctions-adjacent arrangement that RIPE's existing framework cannot address.
Chapter 9: The Cascade Effect โ From Database to Door-Kick
Geographic trust doesn't stop at databases. It cascades through every system that touches IP addresses:
9.1 Law Enforcement: Wrong Door, Wrong Country
In 2016, MaxMind made a decision that would make national news: for any IP address it couldn't geolocate precisely, it assigned the geographic center of the United States โ a farm in Potwin, Kansas, owned by Joyce Taylor. For years, she was visited by FBI agents, IRS investigators, and local police โ all looking for scammers, fraudsters, and criminals whose IP addresses MaxMind had assigned to her coordinates.
This was one database, one default value, one farm. Our data shows 1,357 IPs where databases actively disagree. Each one is a potential wrong door, wrong country, wrong investigation.
๐ Documented Cases of GeoIP Misidentification
- 2016: The Kansas Farm โ MaxMind's default US coordinates led to years of police visits to an innocent family (Ars Technica)
- 2017: IP-Based Raids โ Multiple cases documented where law enforcement used GeoIP data to obtain warrants for wrong addresses (EFF documentation)
- 2020: Tor Exit Node Raids โ Operators of Tor exit nodes raided by police who traced criminal traffic to the exit node's IP, not the actual perpetrator's IP
- 2023: Cloud Provider Warrants โ Legal scholarship documents how cloud infrastructure GeoIP mismatches lead to warrants served in wrong jurisdictions, requiring MLAT processes that can delay investigations by years
9.2 Compliance: Sanctions Screens That Don't Screen
OFAC compliance systems โ used by banks, payment processors, and trading platforms โ use GeoIP to screen transactions. If a transaction appears to come from a sanctioned country, it's blocked. If it appears to come from an unsanctioned country, it's allowed.
Our data shows exactly how this breaks:
- 39 Iraqi IPs appear as UAE โ Iraq is subject to limited US sanctions; UAE is a major financial partner. The misclassification creates a sanctions screening gap
- 8 Russian IPs appear as Polish โ After 2022 comprehensive sanctions against Russia, Polish-classified Russian traffic bypasses screening
- M247's virtual locations (documented in 023C) create explicit sanctioned-country endpoints that GeoIP databases accept without verification
9.3 Content Licensing: Geographic Fiction Costs Billions
The global streaming industry licenses content by territory. Netflix pays different rates for content in different countries. When GeoIP says an IP is in the US, the user gets US content and Netflix pays US licensing fees. When that IP is actually in Singapore (Tencent Cloud's "US" servers), the content is being streamed in the wrong territory โ a breach of licensing agreements that the streaming service doesn't know about because the GeoIP database told them the user was in the US.
The VPN industry โ estimated at $44.6 billion in 2024 and projected to reach $134.35 billion by 2032 (Fortune Business Insights) โ is built entirely on the ability to present a different geographic identity to content providers. The GeoIP databases that enable this geographic fiction are simultaneously the databases that content providers rely on to enforce their territorial restrictions.
๐ The Trust Chain Paradox
The same MaxMind database is used by:
- VPN providers โ who use it to verify their virtual locations "work" (appear as the right country)
- Streaming services โ who use it to block VPN traffic and enforce territorial restrictions
- Compliance systems โ who use it to screen for sanctioned-country traffic
- Law enforcement โ who use it to determine jurisdiction for investigations
- Advertising networks โ who use it to target ads by geography
All five consumers trust the same database. The VPN provider's business model is to defeat that database. The streaming service's business model is to trust that database. When VPN providers succeed (by feeding geofeeds, by getting virtual locations classified correctly), streaming services fail โ and so does every other consumer of the same data. The trust chain is a zero-sum game where one participant's success is everyone else's failure.
Chapter 10: The GeoIP Industry โ Who Profits From Geographic Fiction
IP geolocation is a multi-billion-dollar industry. The major players:
| Provider | Revenue Model | Accuracy Claim | Verification | Market Position |
|---|---|---|---|---|
| MaxMind | GeoLite2 (free) + GeoIP2 (paid) | 99.8% country-level | None published | Market leader, de facto standard |
| IPinfo | API subscriptions ($0-$999/mo) | "Reliable, accurate" | None published | Developer-focused, privacy detection |
| IP2Location | Database downloads + API | "99.5% country accuracy" | None published | Malaysian company, Asia-Pacific focus |
| DB-IP | Free + paid tiers | "Continuously improved" | None published | EU-based alternative |
| Digital Element | Enterprise licensing | "Gold standard" | None published | Acquired by Digital Envoy |
Notice the "Verification" column. Not a single major GeoIP provider publishes independent accuracy verification. Their accuracy claims are self-reported. MaxMind claims 99.8% country-level accuracy โ but our data shows 31.4% of IPs have BGP routing that disagrees with GeoIP classification. These numbers can both be "correct" because they measure different things โ but the consumer doesn't know that.
๐ HIGH FINDING: Unverifiable Accuracy Claims
The GeoIP industry markets accuracy percentages that are self-measured against their own ground truth. No independent audit body verifies GeoIP accuracy. No regulatory standard defines what "99.8% country accuracy" means. No consequence exists for being wrong. The industry profits from the perception of accuracy while being structurally incapable of achieving it for the IP addresses that matter most โ those deliberately designed to mislead.
The Feedback Loop
GeoIP databases don't independently verify locations. They copy from each other:
This is why AbuseIPDB agrees 100% with our derived country โ it's reading from the same source. When M247 publishes a geofeed saying 62.133.46.0/24 is in Tehran, every database in this chain eventually says "Iran." And when compliance systems check three different databases and they all say "Iran," they treat it as high-confidence โ not realizing all three databases trace back to a single, unverified CSV file on a Romanian web server.
Chapter 11: The VPN Review Industry โ Conflict of Interest at Scale
If you search "best VPN" on Google, the top results are review sites. Many of these review sites are owned by VPN companies โ creating a closed loop of self-promotion:
๐ Kape Technologies: The Malware-to-VPN Pipeline
Kape Technologies (LSE: KAPE, formerly Crossrider) was an Israeli-founded company that developed browser extensions classified as malware and adware by antivirus vendors. In 2017-2018, it pivoted to VPN services through acquisitions:
- 2017: Acquired CyberGhost VPN for $10.4M
- 2018: Acquired Zenmate VPN for $5.5M
- 2019: Acquired Private Internet Access (PIA) for $95.5M
- 2021: Acquired ExpressVPN for $936M โ the largest VPN acquisition in history
- Also acquired: vpnMentor and Wizcase โ two of the internet's most-visited VPN review sites
Kape Technologies โ a former malware company โ now owns four of the world's most popular VPN services AND two of the most-trusted VPN review sites. When vpnMentor recommends ExpressVPN as "best VPN," it's recommending its own parent company's product. In 2023, Kape was taken private by Unikmind for approximately $1.6 billion, removing even the minimal transparency that public listing required.
Kape's founder, Teddy Sagi, is an Israeli billionaire who was convicted of securities fraud in 1996 and served a prison sentence. He later built Playtech (online gambling software) into a ยฃ3B company before founding Crossrider/Kape.
๐ Conspirative Question: The Intelligence Value of VPN Ownership
A company that:
- Was founded by a convicted financial criminal
- Previously distributed malware that intercepted browser traffic
- Now owns four major VPN services handling millions of users' traffic
- Also owns the review sites that funnel users to those VPN services
- Was taken private, eliminating public oversight
Has access to the internet traffic of millions of privacy-conscious users who specifically chose VPN services because they wanted privacy. The traffic that VPN users are trying to hide โ from governments, from ISPs, from surveillance โ now passes through servers owned by a private company with a history of intercepting browser traffic, founded by a convicted criminal, with no public reporting requirements.
This is not speculation about what Kape does with this access. It is a structural observation about what Kape can access โ and the uncomfortable reality that the VPN industry's trust model requires users to trust the VPN provider more than they trust their ISP or government. When the VPN provider has a worse track record than either, the trust model collapses.
Chapter 12: The "No-Logs" Lie โ Verified Failures
VPN providers market "no-log" policies as their primary trust guarantee. Several have been proven false:
| VPN Provider | Claim | Reality | Source |
|---|---|---|---|
| PureVPN | "No logs whatsoever" | FBI obtained user logs in 2017 cyberstalking case (US v. Ryan Lin) | BleepingComputer |
| IPVanish | "Zero logs policy" | Provided logs to DHS/HSI in 2016 child exploitation case | TorrentFreak |
| HideMyAss | "No logging" | Provided connection logs leading to LulzSec hacker arrest (Cody Kretsinger, 2011) | Ars Technica |
| EarthVPN | "No logs" | Dutch police obtained customer data via datacenter operator logs (2014) | TorrentFreak |
The pattern: every "no-log" VPN provider that has been tested by law enforcement has produced logs. The providers that haven't produced logs haven't been tested โ not because they don't have logs, but because no court has ordered them to produce any.
๐ Reading Between the Lines
The "no-log" marketing works because most users will never be investigated. For the threat actors in our honeypot database โ the SSH brute-forcers, the botnet operators, the malware distributors โ VPN "no-log" policies are not a privacy guarantee. They are a speed bump. Law enforcement can obtain logs through court orders, national security letters, or MLAT requests. The VPN provider can comply, delay, or relocate to a jurisdiction that makes compliance harder โ but the logs exist because they must (for abuse management, billing, and capacity planning) even if the company claims they don't.
The real protection is not "no logs." It is the cost of obtaining them. A VPN provider in Panama requires an international legal request. A VPN provider in Romania requires an EU mutual legal assistance procedure. The log exists in both cases โ but the effort to retrieve it may exceed the value of the investigation. This is not privacy. It is friction.
Chapter 13: The Discrepancy Countries โ Where Geography Lies Most
Not all countries are equally affected by geographic discrepancies. The countries that appear most often as the "derived" country for discrepant IPs:
| Country | Discrepant IPs | As % of Country's Total | Pattern |
|---|---|---|---|
| ๐ณ๐ฑ Netherlands | 80 | ~15% | Hosting hub โ many foreign entities use Dutch infrastructure |
| ๐ซ๐ท France | 63 | ~12% | OVH effect โ French company with global datacenter presence |
| ๐ฉ๐ช Germany | 50 | ~10% | Hetzner/IONOS/Contabo โ German hosting, international clients |
| ๐บ๐ธ United States | 36 | ~3% | Cloud providers with US RDAP, global routing |
| ๐ธ๐ฌ Singapore | 34 | ~8% | Tencent Cloud hub โ SG routing, worldwide sales |
| ๐ฌ๐ง United Kingdom | 22 | ~5% | M247/VPN infrastructure hub |
| ๐ท๐ด Romania | 20 | ~6% | M247 Europe SRL โ virtual location manufacturer |
| ๐ฎ๐ณ India | 13 | ~4% | Tencent/cloud virtual India endpoints |
| ๐น๐ท Turkey | 13 | ~7% | Turkey/Tรผrkiye naming discrepancy + transit routing |
| ๐จ๐ญ Switzerland | 12 | ~8% | Privacy hosting โ ProtonVPN/Swiss jurisdiction |
| ๐ท๐บ Russia | 11 | ~3% | Post-sanctions geographic obfuscation |
| ๐ง๐ช Belgium | 11 | ~9% | EU institution hosting cross-border effects |
The Netherlands, France, and Germany โ Europe's three largest hosting markets โ produce the most discrepancies. This isn't because these countries are dishonest. It's because they host the most international infrastructure, and GeoIP databases must choose whether a server in Amsterdam owned by a French company serving Brazilian customers is "Dutch," "French," or "Brazilian."
The answer depends on who's asking. GeoIP databases give one answer and pretend it's the only answer.
Chapter 14: The Structural Failure โ Why This Can't Be Fixed
The geographic trust chain is broken at a fundamental level. Not because anyone made a mistake, but because IP addresses don't have geography.
An IP address is a number. It's assigned by a registry (IANA โ RIR โ LIR โ operator), routed by BGP, and hosted on hardware in a physical location. But the number itself has no inherent geographic property. All geolocation is inference โ and every inference method has failure modes:
- BGP routing tells you where packets go โ but anycast, CDN, and transit routing mean packets can take any path
- RDAP registration tells you who has the legal right to the address โ but companies register in jurisdictions of convenience, not geography
- Geofeed self-publication tells you what the operator claims โ with no verification
- Latency measurement tells you how far away the server is โ but CDN edge servers and VPN tunnels break this
- DNS hostname sometimes encodes location โ but operators can name servers anything they want
Every method can be gamed, every method has legitimate failure cases, and no method can distinguish between a gaming and a failure. This is not a bug. It is the architecture of the internet, which was designed for packet delivery, not geographic accountability.
๐ด CRITICAL FINDING: The Unfixable Architecture
The internet's geographic trust problem cannot be solved within the current architecture because:
- No single source of truth exists for IP geolocation
- Legitimate use cases (cloud hosting, CDN, anycast) produce the same patterns as deliberate obfuscation
- RFC 8805 geofeeds institutionalized self-published geography with no verification
- Commercial incentives (VPN services, virtual locations) profit from geographic fiction
- No regulatory body has jurisdiction over IP geolocation accuracy across borders
- GeoIP databases are commercial products, not standards โ they answer to customers, not to accuracy
The trust chain will continue to produce wrong answers for the IPs that matter most โ because the IPs that matter most are the ones whose operators have the strongest incentive to lie.
Chapter 15: Implications for This Investigation Series
This dossier (023D) is part of the VPN Laundromat series that began by asking a simple question: why do our honeypot attackers use VPN services?
- 023A established that VPN providers serve as laundering infrastructure for attack campaigns
- 023B traced the backbone to M247 Europe SRL and its corporate parentage under Macquarie Group
- 023C discovered M247's geofeed manufacturing virtual endpoints in sanctioned nations
- 023D (this dossier) demonstrates that geographic fabrication is not unique to M247 โ it is systemic across 1,357 IPs, affecting 17.1% of our database
The progression is important: what began as an investigation into VPN abuse became an investigation into hosting infrastructure, which became an investigation into geographic fabrication, which became an investigation into the internet's geographic trust model itself.
Each layer revealed a deeper structural problem. The VPN providers aren't the disease โ they're a symptom. The hosting companies aren't the disease โ they're the infrastructure. The GeoIP databases aren't the disease โ they're the transmission vector. The disease is that the internet has no mechanism for geographic truth, and every institution that needs geographic truth โ from OFAC to Netflix to the FBI โ has built their systems on the assumption that one exists.
๐ Reading Between the Lines
There is an industry-wide conspiracy of silence about GeoIP accuracy. MaxMind knows its data is wrong for VPN and cloud infrastructure โ it publishes an IP correction form for users who find errors, implicitly acknowledging the problem. IPinfo knows its privacy detection misses discrepant IPs โ but flagging them would annoy its cloud provider customers. AbuseIPDB knows it's just echoing MaxMind โ but disclosing this would undermine its value proposition.
Nobody lies. Nobody tells the whole truth. Everyone profits from the ambiguity. And every downstream system that depends on geographic accuracy โ compliance, law enforcement, content licensing, advertising โ absorbs the uncertainty without knowing it exists.
This is not a conspiracy in the traditional sense. No one coordinated it. It emerged from market incentives: accuracy is expensive, inaccuracy is free, and the people harmed by inaccuracy (the Joyce Taylors, the wrongly-raided Tor operators, the sanctions-evading VPN users) are not the people paying for the databases.
Chapter 16: Methodology
๐ Data Sources and Methods
- Primary database: LSN Threat Intelligence Platform โ 7,946 IPs enriched from honeypot, Shodan, AbuseIPDB, GreyNoise, RDAP, BGP/Cymru, IPinfo, and 8 additional sources
- Discrepancy identification: Cross-comparison of country fields from BGP/Cymru (routing), RDAP (registration), AbuseIPDB (community), IPinfo (commercial), and Shodan (scanning) โ flagging any IP where two or more sources disagree on country
- Corridor analysis: Grouping discrepant IPs by BGPโRDAP country pairs to identify systematic patterns
- Organization analysis: Grouping by RDAP organization to identify which entities produce most discrepancies
- Multi-source matrix: Comparing all 5+ source opinions for high-threat (โฅ85) discrepant IPs
- IPinfo privacy flags: Cross-referencing deep_ipinfo VPN/proxy/hosting/tor boolean flags against discrepancy set
- OSINT research: Published reports on GeoIP accuracy, VPN industry structure, compliance system design, law enforcement geolocation usage
- Prior series dossiers: 023A (VPN attack laundering), 023B (M247 backbone), 023C (geofeed sanctions)
Chapter 17: Conclusions
The internet's geographic trust chain is broken โ not at one point, but at every point.
๐ด Summary of Findings
- 1,357 IPs (17.1%) in our database have country discrepancies between authoritative sources
- BGP routing agrees with GeoIP only 68.6% of the time โ meaning a third of the internet's physical routing contradicts the geographic claims in databases
- IPinfo flags zero discrepant IPs as VPN/proxy/hosting โ the commercial privacy detection product is blind to the problem
- AbuseIPDB agrees 100% with derived country โ not because it independently confirms, but because it is the primary source. The confirmation is circular
- Major cloud providers (Tencent: 76 IPs, M247: 30 IPs, OVH: 22 IPs) systematically produce geographic fiction through legitimate cloud infrastructure
- Conflict zone traffic (Iraq via UAE: 39 IPs) is systematically misattributed, making the most sensitive traffic the hardest to correctly geolocate
- The VPN review industry is controlled by VPN companies (Kape Technologies owns ExpressVPN + PIA + CyberGhost + vpnMentor + Wizcase)
- Every "no-log" VPN tested by law enforcement has produced logs
- No GeoIP provider publishes independently verified accuracy data
- The problem is architectural โ IP addresses don't inherently have geography, and no proposed solution addresses the fundamental disconnect
"The internet doesn't have a map problem. It has a truth problem โ and geography was just the first place where the lie became measurable."
TI-2026-023D ยท The VPN Laundromat Series ยท Published June 2026
LSN Threat Intelligence ยท shuffle-on.com/threat-intel