Seven Japanese government bodies, led by the Financial Services Agency and the National Police Agency, sent near-identical letters on August 7, 2026 to Google, Meta, TikTok's Japan unit, X Corp. and LINE Yahoo, asking each to tighten checks on advertisers after deepfake ads impersonating celebrities helped drive a wave of social-media investment fraud. The request is formal administrative guidance rather than a legal order: the letters state plainly that it falls under Article 2, Item 6 of the Administrative Procedure Act and is not a disposition carrying penalties.
The trigger, per the letters: ads using AI-generated likenesses and voices of well-known people to promote fake investment schemes are spreading on social platforms, and the resulting SNS-based investment scams cause both financial losses to viewers and reputational harm to the people being impersonated. The letters went out addressed among others to LINE Yahoo chief executive Takeshi Idezawa, Meta chief executive Mark Zuckerberg and X Corp. president Elon Musk, alongside Google and TikTok's Japan entities.
Three asks, three deadlines
The guidance sets out three requirements, each with its own paper trail to the Digital Agency, which is coordinating the effort.
First, advertiser identity verification. Platforms are asked to reliably carry out identity checks on advertisers, including electronic authentication such as Japan's My Number Card for individuals or commercial-registry e-certificates for companies. They must report by October 16, 2026 exactly how and when they run these checks, then follow up by March 16, 2027 with results covering October 2026 through February 2027: verification counts and rates broken down by method and by domestic versus overseas advertisers, plus how many ad listings were rejected for failed identity checks.
Second, ad transparency. Platforms are asked to let users see who is behind a verified ad, including the advertiser's name, location and whether identity checks were completed, along with whether the ad is flagged as an ad, made substantially with generative AI, or targeted for a stated reason. Implementation plans are due October 16, 2026, with results due March 16, 2027.
Third, takedown response. Platforms are asked to act without delay on removal requests for ads that may violate the Penal Code or the Financial Instruments and Exchange Act, whether the request comes from Japan's Internet Hotline Center or a ministry with jurisdiction. They must also report by October 16 how their ad-review standards address impersonation fraud ads, then report removal counts by cause and trigger by March 16, 2027, including the ratio of removals to requests and average response time by requesting body.
The legal hooks are explicit. Using a celebrity's name without permission to fabricate a fake investment track record and circulating it online can breach Article 161(1) of the Penal Code, the government notes, and unregistered operators who lure viewers from an innocuous-looking ad into what becomes securities business can fall foul of Article 29 of the Financial Instruments and Exchange Act. To back that up, the Internet Hotline Center revised its operating guideline on August 7, 2026, and the Ministry of Internal Affairs and Communications revised its guideline under Article 26 of the provider-liability law on August 6, 2026, both now listing impersonation fraud ads as illegal content.
Why YouTube, Instagram and TikTok
The police data behind the request narrows the target list. Among the banner-style ad tools used as the initial point of contact in SNS-based investment fraud cases the National Police Agency tracked for 2025, YouTube accounted for the largest share of those contacts, ahead of Instagram, TikTok, Facebook, LINE and X.
| Platform | Share of Contacts |
|---|---|
| YouTube | 27.1% |
| 16.3% | |
| TikTok | 10.4% |
| 6.5% | |
| LINE | 4.8% |
| X | 2.1% |
None of this carries a fine or license risk on its own. The letters are voluntary requests, and the government's own footnote says so twice. What changes the calculus for the five companies is the paper trail: two rounds of written reporting to the Digital Agency, addressed to a named contact inbox, with specific counts of ads rejected, ads removed and how long each removal took. A platform that ignores the request keeps its legal exposure exactly where it was. A platform that reports thin numbers now has those figures on file with seven agencies at once, due by March 16, 2027, ahead of any decision on whether this guidance hardens into something with more teeth.