Incident sheet · SIG-5591 · /security
- Signal
- SIG-5591
- System
- SEC · Security & Risk
- Filed
- Read
- 4 min · 851 words
- Severity
- Routine
Japan Flags Privacy Risks in Google, Meta, TikTok and LINE Yahoo AI Practices
Japan's 2026 monitoring report finds ad opt-outs at Google, Meta, TikTok and LINE Yahoo do not stop data collection, and demands continued scrutiny of AI training and chatbot ad targeting.

Event log
- Japan's User Information Working Group finalised its 2026 report after deliberations from October 19 to 21, 2026.
- For all four companies, opting out of targeted advertising does not stop the collection or transmission of user data.
- Meta uses Meta AI conversations of users aged 18+ to optimise ads; LINE Yahoo shows ads based on AI prompts with no opt-out.
- Only Google and Yahoo! Japan let people without accounts opt out of ad targeting from externally collected data.
- Monitoring proceeds under Article 52(2) of Japan's telecommunications privacy guidelines, with continued review of AI training and cookie alternatives.
Japan's Ministry of Internal Affairs and Communications has ordered continued scrutiny of how Google, Meta, TikTok and LINE Yahoo handle personal information, after a 2026 monitoring report found that opting out of targeted advertising does not stop any of the four companies from collecting data in the first place.
The report, produced by the ministry's User Information Working Group following deliberations from October 19 to 21, 2026, examined conversational AI, targeted advertising and tracking tools embedded in third-party websites. The working group stated plainly: "It is necessary to check at an early stage" how conversational AI handles personal information for training and advertising.
How do the four platforms use AI conversations for ads?
The differences are sharp. Meta confirmed it uses conversations with Meta AI, for users aged 18 and above, to improve advertising — although it shows no ads inside the chatbot itself. The ministry noted Meta "has made clear that it uses the content of exchanges with Meta AI to optimise advertising."
LINE Yahoo goes further. It already displays ads based on users' AI prompts and, as of August 2026, was considering using earlier conversations for advertising. Users cannot opt out of prompt-based advertising.
Google does not use Gemini conversations for targeted ads, though its conversational AI displayed search-related advertisements as of September. TikTok reported no conversational AI advertising at all.
The takeaway: two of the four platforms already mine chatbot exchanges for ad targeting, and Japan wants that practice watched closely.
What feeds AI training data?
The report mapped each company's training inputs:
- Google: Gemini prompts, uploaded files, feedback and linked-app information, plus third-party information "generally," subject to personal-data considerations.
- Meta: Posts, photographs, captions, Meta AI conversations and third-party information.
- LINE Yahoo: No use of user information in the conversational AI training reviewed.
- TikTok: User-generated content, including for content moderation and ad optimisation.
Opt-out mechanics also diverge. Google requires users to sign in and disable activity settings. TikTok offers a choice. Meta places its controls outside the AI interface — a friction point the ministry flagged.
Profiling splits the group too: "all four use AI technology to optimise advertising, while all except TikTok use it to profile users."
Where does the secrecy of communications come in?
The review raised a question with legal weight under Japanese telecommunications law: is one person's consent enough when AI analyses conversations involving other people, and does consent given years earlier still mean anything? LINE Yahoo said it seeks specific consent when users first activate such features; Google notifies users before enabling smart features. The ministry wants "appropriate explanations to users and technical measures at each company" monitored continuously.
Why do ad opt-outs fall short?
This is the report's core finding for enterprises tracking data flows. "For all four companies, opting out does not stop the transmission of information itself." Data keeps flowing from tracking tools on third-party websites, is not automatically deleted after an opt-out, and "may be used for purposes other than advertising within the scope of their privacy policies."
Deletion routes differ. Google and Meta offer separate deletion options. TikTok lets users remove certain information collected outside its platform. LINE Yahoo stops transmitting location information when users opt out of collection.
Who can opt out without an account?
All four collect device-related information from logged-out users, but only Google and Yahoo! Japan let people without accounts opt out of ad targeting based on externally collected data. TikTok said it does not connect logged-out data to accounts — but its privacy policy "does not state the differences depending on whether someone has an account or is logged in."
At registration, Google and Meta offer an advertising choice; LINE Yahoo does not, and TikTok defers settings to later. LINE Yahoo, Meta and TikTok "do not explicitly explain at account registration that settings can be changed afterwards."
What about transparency and tracking tools?
All four collect device identifiers, advertising identifiers and behavioural data through third-party websites. Google, Meta and TikTok receive hashed email addresses or phone numbers to match ad data; LINE Yahoo converts information into identifiers after receipt. Google and Meta reported corrective action against improper partner practices, but the ministry found "the actual state of correction is unclear" and recommended tracking disclosures be provided "in a list." Google, Meta and TikTok restrict partners from supplying data on children under 13; LINE Yahoo reported no equivalent restriction.
Only Meta measures — partially — whether users understand its privacy policies, having surveyed users in 2022 without testing its later rewrite. The other three do not assess understanding at all.
What happens next?
The working group recommended "continuous monitoring" of advertising, AI training, profiling, cookie alternatives and safeguards for people without accounts. The review operates under Article 52(2) of Japan's guidelines on protecting personal information in telecommunications businesses. The ministry also pressed for domestic privacy staff: "securing domestic personnel and establishing systems are desirable," with measures that "accord with Japan's legal system."
Japan has put all four platforms on notice — the monitoring will not stop.
via medianama.com (Source)
More from Amara Osei
Show full bio
Correspondent covering regulation, policy and their impact on enterprise information technology and the leaders who run at CIO Monitor.
122 articles
The newsroom digest
One email a day with the stories that matter. No noise.
Cross-references
- OpenAI’s Business Hub makes ChatGPT more search-like
- Internal Memo Reveals Meta's Plan to Kickstart Enterprise AI Business
- OpenAI's Businesses in ChatGPT Takes Aim at Salesforce, SAP and Google
- Alphabet Loses $500 Billion in a Day as AI Search Erodes Ad Value
- Dialpad Swallows Ada for $430M and Resets the Conversational AI Market