Skip to main content
AVOID.NET
← AI Dating Scam Network (Dora / Doni / Romi cluster)reviewed 2026-10-28 · 24 claims checked

Fact-check findings

What an automated fact-checker found when it re-read AI Dating Scam Network (Dora / Doni / Romi cluster) against the sources the page cites. Only the most recent review is shown.

Read this first

These findings are produced by an automated reviewer, and its results vary between runs: the same page, checked three times on the same day, came back with 15%, 20% and 34% of its claims disputed, mostly because each run extracted a different number of claims. Treat what follows as leads, not rulings.

“Disputed” means the reviewer could not reconcile the claim with the evidence it cited. It does not mean the claim is false. “Unverifiable” means no reachable source settled it either way.

Nothing here changes the page on its own. A proposed correction is applied only after a human moderator approves it; until then the page reads as it did when reviewed.

disputed 2unverifiable 5link rot 4partially supported 2confirmed 116 corrections pending · 0 applied

disputed

2 claims

The reviewer could not reconcile the claim with the evidence it cited. This is a lead, not a ruling that the claim is false.

  1. #5[disputed][awaiting moderator]in section: Overview and Identity
    “The apps' internal documentation, code repositories, and analytics services were hosted on Chinese infrastructure including Tencent Cloud, Feishu, Gitee, and ByteDance analytics, according to reporting by The Daily Guardian citing Anthropic's September 2026 threat report.”
    reviewerThe apps' internal documentation, code repositories, and analytics services were hosted on Chinese infrastructure including Tencent Cloud, Feishu, Gitee, and ByteDance analytics.The only outlet cited for this specific infrastructure detail is now a dead link, and the primary Anthropic report I was able to read directly describes different infrastructure (Google Cloud backend, PRC-based proxy/reseller), not Tencent Cloud/Feishu/Gitee/ByteDance.
    Proposed correction (not yet applied)
    The apps' backend was reportedly hosted on Google Cloud, with the operation relying on PRC-based API reseller and proxy infrastructure to access AI providers, per Anthropic's September 2026 threat report; the Tencent Cloud/Feishu/Gitee/ByteDance attribution previously reported by The Daily Guardian could not be confirmed against Anthropic's published GTG-15001 case study, which instead names a Google Cloud-hosted backend and PRC-based proxy/reseller infrastructure.
    citeddailyguardian.ca/the-sexy-ai-powered-dat…-scams-are-here/www.anthropic.com/threat-intelligence-re…t-september-2026
  2. #22[disputed][awaiting moderator]in section: Known Infrastructure and Attribution
    “Anthropic attributed the operation to a China-based app studio based on the operator's use of Chinese-language internal documentation, Chinese cloud and software infrastructure (Tencent Cloud, Feishu, Gitee, ByteDance analytics), and the geographic targeting pattern — the apps did not function within China and monetization was disabled in most Asian markets.”
    reviewerAnthropic attributed the operation to a China-based studio based on Chinese-language documentation, Chinese cloud/software infrastructure (Tencent Cloud, Feishu, Gitee, ByteDance analytics), and geographic targeting patterns.Same underlying infrastructure-attribution issue as the Overview section; repeated here with the same unsupported specifics, and now contradicted by the primary source's own described infrastructure.
    Proposed correction (not yet applied)
    Anthropic attributed the operation to a China-based app studio based on the operator's use of Chinese-language internal documentation and its reliance on PRC-based API reseller and proxy infrastructure to access AI providers; Anthropic's published report describes the backend itself as hosted on Google Cloud and does not name Tencent Cloud, Feishu, Gitee, or ByteDance analytics in connection with this case.
    citedtechjournal.org/claude-dating-app-scamdailyguardian.ca/the-sexy-ai-powered-dat…-scams-are-here/www.ntd.com/chinese-dating-apps-target-a…pic_1172452.html

unverifiable

5 claims

No source the reviewer could reach confirms or contradicts the claim.

  1. #6[unverifiable][awaiting moderator]in section: Overview and Identity
    “The apps did not function within China, and monetization was reportedly disabled across most Asian markets outside China, per the same reporting.”
    reviewerThe apps did not function within China, and monetization was disabled across most Asian markets outside China.Only source for this specific claim is dead, and I could not corroborate it elsewhere, including in the primary Anthropic report.
    citeddailyguardian.ca/the-sexy-ai-powered-dat…-scams-are-here/
  2. #13[unverifiable][awaiting moderator]in section: App Store Takedown Status
    “Specifically, Dora, Jovia, Romi, Luma, and a variant called Eterna were reportedly removed from Google Play by approximately September 3, 2026, and Nalo by approximately September 7, 2026.”
    reviewerSpecifically, Dora, Jovia, Romi, Luma, and a variant called Eterna were reportedly removed from Google Play by approximately September 3, 2026, and Nalo by approximately September 7, 2026.The cited source does not contain these specific per-app removal dates; I could not find them in any other source searched. The claim appears to be more precise than the evidence supports.
    citedonlinepersonalswatch.com/2026/09/ai-dati…d-google-stores/
  3. #14[unverifiable][awaiting moderator]in the timeline
    “2026-09-03”
    reviewerDora, Jovia, Romi, Luma, and Eterna reportedly removed from Google Play Store on approximately September 3, 2026.Same underlying issue as the section 5 finding; the cited source does not support this specific date.
    citedonlinepersonalswatch.com/2026/09/ai-dati…d-google-stores/
  4. #15[unverifiable][awaiting moderator]in the timeline
    “2026-09-07”
    reviewerNalo reportedly removed from Google Play Store on approximately September 7, 2026.Same underlying issue as the section 5 and timeline[2] findings; the cited source does not support this specific date.
    citedonlinepersonalswatch.com/2026/09/ai-dati…d-google-stores/
  5. #23[unverifiable][awaiting moderator]in section: Known Infrastructure and Attribution
    “Anthropic attributed the operation to a China-based app studio based on the operator's use of Chinese-language internal documentation, Chinese cloud and software infrastructure (Tencent Cloud, Feishu, Gitee, ByteDance analytics), and the geographic targeting pattern — the apps did not function within China and monetization was disabled in most Asian markets.”
    reviewerThe apps did not function within China and monetization was disabled in most Asian markets (Known Infrastructure section).Same underlying sourcing gap as the Overview section's equivalent claim.
    citeddailyguardian.ca/the-sexy-ai-powered-dat…-scams-are-here/

partially supported

2 claims

The cited evidence supports part of the claim but not all of it.

  1. #11[partially supported][awaiting moderator]in section: Operational Mechanics and Payment Model
    “Gig workers were reportedly compensated on a per-message, per-call, and per-social-media-follow basis and could cash out at low threshold amounts.”
    reviewerGig workers were reportedly compensated on a per-message, per-call, and per-social-media-follow basis and could cash out at low threshold amounts.Compensation structure is confirmed; the 'low threshold amounts' cash-out detail could not be found in any source I could access.
    citeddailyguardian.ca/the-sexy-ai-powered-dat…-scams-are-here/www.progressiverobot.com/2026/09/17/ai-p…-claude-catfish/
  2. #18[partially supported][awaiting moderator]in section: Pig-Butchering Parallels and Fraud Taxonomy
    “Security researchers and journalists covering the scheme have noted structural similarities to pig-butchering romance scams, a fraud category that typically involves prolonged trust-building through romantic interaction before extracting large payments — often via cryptocurrency — from victims.”
    reviewerSecurity researchers and journalists covering the scheme have noted structural similarities to pig-butchering romance scams.None of the three cited sources actually frame this specific network as pig-butchering; the attribution to named 'security researchers and journalists covering the scheme' making that comparison is not supported.
    citedwww.progressiverobot.com/2026/09/17/ai-p…-claude-catfish/www.cryptotimes.io/2026/05/05/2026-pig-b…scam-crackdowns/www.analyticsinsight.net/artificial-inte…ng-romance-scams

confirmed

11 claims

The cited evidence supports the claim as written.

  1. #1[confirmed][no action needed]in the summary
    “A China-based app studio, tracked by Anthropic as GTG-15001, operated a network of more than 20 fraudulent dating applications that allegedly used Anthropic's Claude AI to power thousands of undisclosed fake personas.”
    reviewerA China-based app studio tracked by Anthropic as GTG-15001 ran a network of more than 20 fraudulent dating apps that used Claude to power undisclosed AI personas.Corroborated directly from Anthropic's report and multiple independent secondary outlets.
    citedwww.anthropic.com/threat-intelligence-re…t-september-2026
  2. #2[confirmed][no action needed]in the summary
    “During a two-week period in April 2026, the operation engaged at least 25,000 users — primarily men in the United States — through approximately 4,700 AI-generated profiles that sent roughly 2.36 million messages while users paid real money via in-app coin purchases to continue conversations.”
    reviewerOver a two-week window in April 2026, roughly 4,700 AI personas engaged at least 25,000 users, primarily US men, sending ~2.36 million messages, with users paying via in-app coin purchases.Figures confirmed by direct read of the Anthropic report and consistent secondary reporting.
    citedwww.anthropic.com/threat-intelligence-re…t-september-2026
  3. #3[confirmed][no action needed]in the summary
    “Anthropic published findings in September 2026, banned associated accounts, and reported the operator to Apple and Google; most named apps were removed by early September 2026, though Kira reportedly remained on Google Play as of September 16, 2026.”
    reviewerAnthropic published findings in September 2026, banned accounts, reported the operator to Apple/Google; most apps removed by early September 2026, Kira remained on Google Play as of Sept 16, 2026.General framing in the summary is appropriately hedged; the overstated specific dates appear only in the body/timeline (see separate findings).
    citedwww.anthropic.com/threat-intelligence-re…t-september-2026onlinepersonalswatch.com/2026/09/ai-dati…d-google-stores/
  4. #4[confirmed][no action needed]in section: Overview and Identity
    “Ten branded app names have been publicly identified: Dora, Doni, Romi, Luma, Jovia, Kira, Gracechat, Haven, Nalo, and Lovia. A later-referenced variant called Eterna also appears in some reports. Anthropic identified two associated Android package identifiers: com.qiga.vio and com.cavalier.nalo.”
    reviewerTen branded app names (Dora, Doni, Romi, Luma, Jovia, Kira, Gracechat, Haven, Nalo, Lovia) have been identified, plus a variant Eterna; two Android package IDs (com.qiga.vio, com.cavalier.nalo) were identified.App name list and package IDs both independently confirmed.
    citedtechjournal.org/claude-dating-app-scamwww.anthropic.com/threat-intelligence-re…t-september-2026
  5. #7[confirmed][no action needed]in section: Scale and Scope of Alleged Fraud
    “According to reporting based on the Anthropic findings, approximately 75 percent of profiles visible in user match feeds were AI-generated; real gig workers accounted for the remaining 25 percent.”
    reviewerApproximately 75% of profiles visible in user match feeds were AI-generated; real gig workers accounted for the remaining 25%.Confirmed directly from the primary Anthropic report text and a secondary outlet.
    citedwww.storyboard18.com/digital/anthropic-u…-ws-l-110424.htm
  6. #8[confirmed][no action needed]in section: Scale and Scope of Alleged Fraud
    “The primary target demographic was reported to be men in their mid-to-late 30s, primarily located in the United States.”
    reviewerThe primary target demographic was men in their mid-to-late 30s, primarily in the United States.Confirmed by direct fetch, though it is the only one of the four cited sources in this section that actually states an age range.
    citedwww.progressiverobot.com/2026/09/17/ai-p…-claude-catfish/
  7. #9[confirmed][no action needed]in section: Scale and Scope of Alleged Fraud
    “That equates to roughly 169,000 messages per day and approximately 94 messages per targeted user during the sample period.”
    reviewerThat equates to roughly 169,000 messages per day and approximately 94 messages per targeted user during the sample period.Both figures are also consistent arithmetic (2.36M/14 days ≈ 169,000; 2.36M/25,000 ≈ 94).
    citedwww.progressiverobot.com/2026/09/17/ai-p…-claude-catfish/
  8. #10[confirmed][no action needed]in section: Operational Mechanics and Payment Model
    “The payment processing was reportedly routed through third-party processors rather than native app store billing, using a hidden in-app browser that could be toggled remotely and concealed during app store review.”
    reviewerPayment processing was routed through third-party processors via a hidden in-app browser that could be toggled remotely and concealed during app store review.Directly confirmed in the cited NTD article.
    citedwww.ntd.com/chinese-dating-apps-target-a…pic_1172452.html
  9. #12[confirmed][no action needed]in section: Anthropic Policy Violation and Response
    “Anthropic's detection was triggered in part by anomalously high API request volumes: the operation generated approximately 100,000 daily API requests, which flagged internal monitoring systems.”
    reviewerAnthropic's detection was triggered in part by anomalously high API request volumes of approximately 100,000 daily API requests.Confirmed via secondary reporting on the researcher's own account of the detection trigger, consistent with the page's figure.
    citedwww.anthropic.com/threat-intelligence-re…t-september-2026www.therundown.ai/news/anthropic-claude-…t-september-2026
  10. #16[confirmed][no action needed]in section: App Store Takedown Status
    “However, Kira — which reportedly shares the same underlying codebase as the removed apps — remained accessible on Google Play as of September 16, 2026, more than a week after Anthropic's public disclosure on September 10, 2026.”
    reviewerKira remained accessible on Google Play as of September 16, 2026, sharing the same codebase as previously removed apps, more than a week after Anthropic's September 10, 2026 disclosure.Confirmed directly. The live Google Play URL cited as a separate source in this section is now dead; see the related link_rot finding.
    citedonlinepersonalswatch.com/2026/09/ai-dati…d-google-stores/
  11. #24[confirmed][no action needed]in section: Known Infrastructure and Attribution
    “All named app variants reportedly shared a common backend codebase, with code shuffling applied across variants to obfuscate the shared origin.”
    reviewerAll named app variants reportedly shared a common backend codebase, with code shuffling applied across variants to obfuscate the shared origin.Corroborated by The Verge's independent technical investigation (security researcher Matthew Gore-Kormanik found the apps 'share code to a T' and use the same backend architecture), which is not cited on the page but supports this claim.
    citedtechjournal.org/claude-dating-app-scamwww.ntd.com/chinese-dating-apps-target-a…pic_1172452.html
How this fits together. The reviewer reads the published page and its cited sources and records one finding per claim. A human moderator decides whether each proposed correction is applied; those decisions, and the score changes they cause, appear in the audit log. Earlier review runs are not shown here; only the latest reflects the page as it stands.