Data from The GDELT Project · Reviewed and interpreted by RSRC
Search the research collection
Find reporting by topic, place or language. Open a record to see why it was included and what the available evidence supports.
The chart uses every search filter above and all matching records, across every results page. The percentage denominator is all included URLs in the same dates, before theme, language, country or text filters.
Distinct included URLs matching these filters in each observation month. Select a point to explore its records. On small screens, scroll the chart horizontally.
A URL counts once per month and can appear in several months or themes. Themes describe the current record assessment; months show URL observations, not when events or each claim first appeared. Warnings, allegations and reported experiences are all included when coded. Partial months, collection gaps and changes in sources affect both measures. A higher share does not establish a higher risk to victims.
| Observation month | Matching URLs | All included URLs | Share |
|---|---|---|---|
| Jan 2025 | 264 | 591 | 44.7% |
| Feb 2025 | 554 | 1,629 | 34.0% |
| Mar 2025 | 141 | 561 | 25.1% |
| Apr 2025 | 116 | 390 | 29.7% |
| May 2025 | 193 | 398 | 48.5% |
| Jun 2025 | 53 | 209 | 25.4% |
| Jul 2025 | 151 | 522 | 28.9% |
| Aug 2025 | 129 | 447 | 28.9% |
| Sep 2025 | 164 | 613 | 26.8% |
| Oct 2025 | 201 | 483 | 41.6% |
| Nov 2025 | 164 | 392 | 41.8% |
| Dec 2025 | 124 | 333 | 37.2% |
| Jan 2026 | 67 | 530 | 12.6% |
| Feb 2026 | 224 | 1,106 | 20.3% |
| Mar 2026 | 90 | 360 | 25.0% |
| Apr 2026 | 145 | 491 | 29.5% |
| May 2026 | 200 | 785 | 25.5% |
| Jun 2026 | 121 | 842 | 14.4% |
| Jul 2026 | 191 | 1,143 | 16.7% |
| Aug 2026 | 224 | 640 | 35.0% |
| Sep 2026 | 139 | 638 | 21.8% |
| Oct 2026 | 6 | 30 | 20.0% |
Each record represents a source web page, not a verified incident. Different URLs can describe the same event. About these records.
Christiane’s app romance is explicitly framed as love scamming followed by a money request. The retained passage establishes a request, not completed payment or a specified false persona.
Reported caseThe source names Frank Abel, aged 31, on trial for allegedly taking EUR 1,450 from a Padua woman after chat contact and a false war-surgeon/Kabul/chest-in-England persona. Kabul and England are scam pretexts, not actual operation locations. Trial is not conviction or sentence.
Reported caseThe Italian source warns about love scams on social platforms and how to avoid deception. The thousands-at-risk language is a general warning, not a measured victim loss, evaluated intervention or country location.
Prevention & supportKaarst reporting says AI kidnapping images were a fake pretext in a senior woman’s love-scam loss, with police warnings. This is not verified kidnapping, trafficking or actual captivity.
Reported caseThe source explicitly describes a romance-and-investment rare-oil scheme using fake female dating profiles and a girlfriend’s photograph, allegedly taking Rs 1.90 lakh (190,000 rupees) from an AIIMS researcher. AIIMS is not evidence of an AI tool; Ugandan nationality does not locate the operation. Keep the source rupee wording; neither the currency abbreviation nor the offender’s nationality assigns an operation or victim country.
Reported caseThe source names Frank Abel, aged 31, on trial for allegedly taking EUR 1,450 from a Padua woman after chat contact and a false war-surgeon/Kabul/chest-in-England persona. Kabul and England are scam pretexts, not actual operation locations. Trial is not conviction or sentence.
Reported caseThe widow/ChatGPT source’s title describes a threatened home and saved funds, while its paragraph says she lost nearly another GBP 1 million and gives an unverified 300 percent holiday-crypto-fraud claim. Preserve inconsistent near-loss/completed-loss wording and threatened housing, not a harmonized actual loss or proven home seizure. AI assistance is not attacker face/voice generation.
Reported caseWildeshausen reporting alleges a recruitment conversation teaching a woman to target older men’s savings through dating apps, followed by a police apartment search finding firearms/ammunition. Planned extraction is not completed victim loss; found weapons are not proof of an additional firearms offence.
Reported caseGrossenkneten police warn about false social/dating profiles after a local incident and mention another case nearing a court conclusion. Keep general prevention/impersonation/platform context; no specific conviction or identity connecting the two cases is established.
Prevention & supportThe second section URL repeats the retired Chen/fake female soldier/TWD 40,000 interception account. Preserve URL and source identity without inflating victim or loss counts.
Reported caseA retired Chen man in his seventies believed in a cross-border romance with a supposed Korean-American female soldier in Syria. Police/bank staff blocked TWD 40,000 for a fabricated airport/customs emergency. The supposed detained partner is a pretext, not verified custody, military affiliation or operating geography; no completed loss.
Reported caseDa Nang police reportedly arrested three South Korean nationals accused of deepfake romance fraud and trafficking victims into forced online fraud. Record source-specific jurisdiction, alleged network/technology and coercion; nationality alone does not establish a Korean operating base.
Reported caseNasushiobara police describe a 46-year-old woman losing crypto worth JPY 11,919,139 after a supposed Japanese man’s contact developed into intimate conversations and investment urged to build trust. Persona nationality is not a geographic role; do not turn relative/yearless dates into full dates. The victim explicitly reported the loss to police; do not infer an arrest from that filing.
Reported caseFrench reporting describes fabricated Ukrainian-soldier identities, prolonged emotional relationships and thousands of euros sent. Preserve general/multiple-case mechanism claims and source-specific French victims where explicit; soldier nationality is a persona, not a verified military or operation location. Social-network contact is explicit in the headline; cite that field rather than infer it from the paragraph.
Research & trendsAuthorities reportedly arrested three South Korean men in Vietnam linked to deepfake romance fraud and forced victims into online fraud operations in Cambodia. Retain the explicit coercion, alleged losses exceeding USD 1 million and country roles; no completed extradition or perpetrator guilt established.
Reported caseNeipu police/store staff stopped a 55-year-old Hu woman buying TWD 60,000 gift cards for a supposed Shanghai husband’s airport-injury settlement. Fabricated relationship and emergency/payment pretexts are explicit. The alleged injured traveller is a pretext, not proved physical harm, and Shanghai/Kaohsiung do not locate the perpetrator.
Reported caseThe Taichung headline says a 60-year-old woman in a fake Doctors Without Borders trap lost TWD 27.72 million, but the description gives TWD 27.75 million over 83 payments. Preserve this discrepancy; five couriers/coin dealers detained and assets/Bitcoin ATMs seized do not prove reimbursement or a laundering conviction. The Doctors Without Borders trap is source-labelled; real affiliation is unverified. Bitcoin ATM seizure is explicit, not a proved communication channel or money returned.
Reported caseVietnamese reporting says three South Korean fugitives accused of deepfake-driven romance fraud were arrested in Da Nang after a cross-border chase. Preserve alleged network/technology and actual reported arrest; nationality is not the operating location and a chase does not establish a completed extradition.
Reported caseThe source describes a 68-year-old retiree losing EUR 240,000, savings and his home to a fake Lara Fabian relationship and receiving consolation from the real singer. Keep financial/housing loss and practical support attributed; the singer is impersonated, not an offender, and clinical outcomes/date are absent. The singer’s consolation is emotional support, not a practical recovery intervention or repayment.
Reported caseTaichung reports a 60-year-old woman in a fake-dating trap paying 83 times through transfers, cash handovers and crypto, losing TWD 27.75 million. Police detained five couriers/coin dealers and seized USDT, Bitcoin ATMs and vehicles. These are 83 payments, not victims; seized assets are not confirmed reimbursement and general ATM laundering explanation is not proof every named person laundered. Bitcoin ATMs are explicitly described, but a romantic communication platform is not. Retain the technical device separately from crypto transfer rather than inventing a platform.
Reported caseKuroishi police report an Aomori man in his twenties losing JPY 2.45 million after dating-app Momoko, Line and JHYD crypto-investment contact. Initial displayed/received profits and later increased payments are attributed; no verified identity, nationality or full year for October/November dates inferred. Small displayed/received profits are described, but a separate explicit return promise is absent. Keep the investment solicitation and omit a duplicate promised-returns subtype.
Reported caseGerman guidance describes an exciting online acquaintance talking about investments and warns of costly love scams. General warning is not a completed victim loss or evaluated advice; access/paywall label does not license full-body retention.
Prevention & supportTaiwan police publish mixed Xiaohongshu fraud statistics and a network-access restriction, explicitly describing fake dating leading to Line investment groups. Retain the 43 fake-dating category as attributed and separate from the 1,706 two-year and other subtype counts. Investment AI videos do not prove AI-assisted romantic impersonation; paid-sex fraud is a separate category. The mixed-fraud paragraph explicitly reports AI-made financial-analysis videos. Code that attributed technique only in mixed scope; it does not establish romantic face/voice impersonation.
Related mentionHachinohe police report an Aomori man in his sixties losing about JPY 1.5 million in SNS romance fraud involving a purported woman Son-a. The retained short source does not detail the online-store mechanism shown by a richer variant; no incident year or nationality is inferred.
Reported caseThe source describes arrest of an upstream suspect and a request for incommunicado detention in the Jingan romance-scam call-centre investigation, discovered while investigating an assault on a female dentist. The assault is contextual to discovery, not established as harm caused by romance fraud. A detention request is not a granted order.
Reported case