Age and Online Risk — What Japan's 2025 Fraud Statistics Show

I started worrying about an older family member and did not know what to look at first. In August 2026 I read three public Japanese sources: the National Police Agency's fraud statistics, the National Consumer Affairs Center's consultation data, and the Ministry of Internal Affairs' survey of internet use.

I expected all three to show the same shape — risk rising steadily with age. None of them did.

The clearest number is one that went down

Japan's National Police Agency publishes annual figures on tokushu sagi, its term for remote impersonation fraud: someone calls or messages a victim, gains their trust without ever meeting them, and has money transferred or handed over.

In the confirmed figures for 2025 (published 22 May 2026), people aged 65 and over accounted for 51.3% of recognized cases. That is down 14.0 points from the previous year.

All figures are shares of recognized cases, excluding corporate victims. Inside a single category, the range runs from 20.9% to 98.9%. Deposit fraud and cash-card theft are almost entirely committed against people 65 and over. Fictitious billing is a third.

The report also explains the drop. A method the agency began counting separately in 2025 — fraud in which the caller impersonates a police officer and claims to be conducting an investigation — spread to younger victims. People in their twenties and thirties made up 27.9% of relative-impersonation cases (up 14.4 points) and 30.0% of fictitious billing cases.

But the count and the money sit in different places. For police-impersonation fraud the largest number of cases was among people in their thirties (2,239 cases), followed by their twenties (1,718). The largest losses were among people in their seventies (¥27.46bn) and sixties (¥25.63bn). Younger victims show up in the case count; older victims show up in the amount.

Consultations are rising, but 38.6% does not mean what it looks like

The National Consumer Affairs Center reports that in fiscal 2024 there were 304,130 consumer consultations where the contracting party was 65 or over (published 3 September 2025). That is 38.6% of all consultations — the highest share since fiscal 2020.

The 38.6% is the share older people make up of all consultations. It is not the share of older people who ran into trouble. Those are different denominators, and the second one is not in this data at all.

The amounts also ran opposite to what I assumed. The average contracted purchase was about ¥900,000 for those under 65 and about ¥710,000 for those 65 and over — lower for the older group. The average amount already paid was about ¥450,000 and ¥460,000 respectively, effectively the same. For the 65-and-over group, more than 60% of contracted purchases were under ¥50,000.

They are online, but they use it for different things

The Ministry of Internal Affairs' 2025 communications usage survey (published 29 May 2026, surveyed end of August 2025) puts individual internet use at 92.0% of men and 90.4% of women aged 60–69, 77.0% and 67.5% for 70–79, and 45.3% and 32.3% for 80 and over.

What they do with it is asked only of respondents who said they used the internet at all, so the percentages below are not shares of everyone in that age group.

Across all internet users, 48.2% reported buying goods or services online. But in the age breakdown of the top five uses, email is first in every bracket from 50 upward, and shopping does not appear in the top five at all from 60 upward.

That was the part I had wrong. People do not drop off the internet as they age. They stay on it, and the centre of gravity of what they do there moves.

The three sources are not looking at the same year

Each is the latest of its kind, but the periods do not line up — the statistics that break out the 65-and-over group are annual and published on different schedules. They cannot be read as describing the same year.

One more caveat: from 2026 the police statistics treat police-impersonation fraud as its own category and fold social-media investment and romance fraud into the impersonation-fraud total. Next year's numbers will not connect cleanly to this year's.

What follows is my reading, not the data

If the kind of risk changes with age rather than simply increasing, then "restrict everything because they are older" is the wrong shape of response. Some methods target almost exclusively people over 65; others are spread across every age.

Meanwhile the cost of checking has been going up on the screen side, for everyone. Countdown timers. Confirmation dialogs that repeat. Reading a code out of a text message and carrying it back to the page you were on. None of that is easy at any age.

So should there be more confirmation steps? I do not think so. If something asks "are you sure?" every time, people stop reading it and press the button. What matters is not the number of checks but stopping once, at the moment the stakes actually rise.

Family cannot sit next to someone forever

Asking a family member when something looks wrong is a good answer. It is just not available every time, and nobody wants to make a phone call before every transaction.

What I would rather build is something that catches the small doubts — the ones too small to phone anyone about — where they happen. Not deciding on someone's behalf. Just making one more opportunity to look, at the point where a mistake would be expensive. Not restricting everything by age, but placing a single small pause where things tend to go wrong.