Detecting SNS Investment Fraud and Impersonation Ads in Japan
A familiar face, a natural-sounding voice or a bank name is not enough to establish that an investment ad is genuine. Reviewers need a way to find questionable material, preserve the evidence and check the advertiser.
Japan’s 2026 figures
The National Police Agency’s October 2 release reported 7,097 SNS investment-fraud cases and JPY 93.74 billion in losses for January–August 2026, up 53.8% year on year. These are provisional figures and a separate category from SNS romance fraud.
Government and platform measures
In a document dated August 7, 2026, seven Japanese government bodies asked Google, LINE Yahoo, Meta, TikTok and X to strengthen measures including advertiser identity checks and ad transparency. A request is not evidence that every measure has been implemented or verified.
Signals in public ads
Check combinations of signals against official communications and registration records.
| Signal | What to check |
|---|---|
| A famous person or financial institution | Does the official site carry the same offer? |
| Guaranteed returns or limited places | Who is advertising, and what supports the claim? |
| Redirection to LINE or another app | Does the destination match the claimed official contact? |
| Similar ads across accounts | Wording, timing, advertiser information and shared destinations |
AI narrows the queue; people verify
A possible workflow uses text or image similarity to find candidates, then a generative model to organize reasons for review. The model’s explanation must be checked against the original. This describes a design option, not a claim about a deployed model or accuracy.
- Preserve URLs, timestamps, images, displayed advertisers and destinations.
- Group similar content and prioritize items for review.
- Check official statements, business registration and original media.
- Record evidence and decisions; refer or report through authorized channels and retain corrections.
Limits and personal information
Public monitoring does not reveal private chats or transaction histories. Legitimate ads and news can share similar wording, and AI explanations may be wrong. Limit personal information, define the purpose and agree retention.
Related methods and services
FAQ
Can AI determine that an ad is fraudulent?
No. Use it to find candidates and organize checks; people verify the original evidence and decide what action is appropriate.
Can public monitoring see private LINE groups?
No. Do not assume access to private conversations or transactions; relevant organizations must use appropriate authority and procedures.
Where does this fit for a bank?
Public impersonation ads and redirection links can provide investigation leads before a payment. They do not replace customer transaction monitoring.
Is JPY 93.74 billion a full-year total?
No. It is the NPA’s provisional SNS investment-fraud loss total for January–August 2026, published on October 2.