Let's talk about something quite thought-provoking that happened recently.

Many of you may have seen the batch of AI digital humans called "Xiaohai" that CCTV exposed. In the footage, a teenage boy from a single-parent family, living in poverty, says he hides in the classroom gnawing on steamed buns and is saving money to buy meat buns for his auntie mom.

Image source: Internet

In the comments section, elderly viewers sent voice message after voice message: "Xiaohai, why don't you go eat? Don't you have money? Auntie mom's heart aches so much for you."

In May this year, railway police in Urumqi found a 79-year-old woman in a train carriage. She was illiterate and could not say exactly where she was going. She only knew she was taking a train to Yuncheng, Shanxi, to visit relatives—and to see a poor child who called her "auntie" every day on her phone.

It was not until the police stopped her that she learned the child did not exist in reality at all. The elderly woman had secretly bought a ticket without telling her family, carrying a simple concern, and nearly traveled thousands of kilometers to see a piece of code.

Nearly half a year has passed since then, but the Xiaohais have not disappeared. They have changed names and are still crying in elderly people's phones, calling them "auntie mom."

01 This Script Is Mass-Produced on an Assembly Line

Xiaohai is not one person; it is a replicable template.

First, use tragic stories to establish a pseudo-kinship relationship, letting elderly people find a sense of value in being needed; then use eating steamed buns and having no money to eat to trigger sympathy; finally, in a tone of caring for auntie mom, recommend products such as ejiao and morel mushrooms.

Every step lands squarely on the point where elderly people crave companionship and fear being forgotten. The threshold for this operation is absurdly low.

For a few yuan, you can buy a full set of AI digital human generation tutorials. Change a face and change a voice, and you can mass-generate hundreds or thousands of Xiaohais.

One account operator said: "If you pick the right track for middle-aged and elderly people, earning two or three thousand a day is on the low end." The production cost of one video is almost zero, and what it targets is the retirement money elderly people have saved their whole lives.

This model is being replicated onto more personas. Today it is the poor boy Xiaohai; tomorrow it might be a filial daughter, a sensible granddaughter, or a son working away from home. The script changes, but the routine does not.

02 A Tiny AI-Generated Label Cannot Stop a Call of "Auntie Mom"

Whenever people raise doubts, the platform defends itself: We labeled it.

Reporters found that in some Xiaohai videos, the lower left corner indeed says the content is AI-generated, and some account bios also add that it is an AI digital human and viewers should be rational.

But for an elderly person who is illiterate and has no idea what AI is, this line of small text is functionally nonexistent. She will not know that "AI-generated" means this person does not exist at all.

Image source: Internet

What is even more helpless is that some netizens specifically posted videos reminding people that Xiaohai is AI, yet some elderly people still did not catch on and treated it as new plot content.

The original purpose of labeling is to remind. When the people being reminded cannot receive this information at all, the line of text is left with only symbolic meaning—for the platform to use to cut off responsibility.

Legally, the platform has indeed fulfilled its duty to inform.

But if the person being informed cannot understand it, that duty becomes an empty piece of paper.

What elderly people need is not a line of gray small text, but a way she can understand, to be told that the person on the screen is not real.

03 Ban One, Another Fills In

After the exposure, accounts such as "Haikai Hui Nulli Deng" were banned, renamed, and had videos deleted, and the actions looked quick.

But when reporters searched, "Haikai Yao Nulli" and "Nulli de Haikai" popped up again, changing names and continuing to perform.

Ban one, and the next mass-generated one immediately fills the spot. The platform's post-event handling is like whack-a-mole, always chasing the scams.

The platform has algorithms, data, and review capabilities.

The same set of boy images appearing repeatedly, highly homogeneous scripts, comment sections full of elderly users expressing sympathy and placing orders—these signals are enough for a high-risk account to be intercepted at the publishing stage.

Image source: Internet

The crux of the problem is not whether it can be identified, but whether the platform is willing to move review from post-event account bans to pre-event filtering.

Some lawyers have proposed that platforms should conduct security assessments of virtual image marketing beforehand and mark it prominently, carry out review and real-time inspections during the process and establish risk identification models, and establish and improve complaint and reporting mechanisms afterward.

Platforms that provide algorithm recommendation services to elderly people should also legally carry out monitoring, identification, and handling of telecom and online fraud-related information. Labeling is only the bottom line, not the endpoint.

04 What Can Be Done to Prevent This?

The most heartbreaking part of this matter is not the technology itself.

That 79-year-old woman is illiterate and cannot tell AI from real people, but she knows that when a child eats steamed buns, her heart aches.

What drove her to get on a train alone was not stupidity; it was loneliness. Scams play the emotional card, so platforms cannot just play the compliance card.

A line of "AI-generated" label cannot block a call of "auntie mom."

And one step earlier than platform review is in each of our own homes.

If you have elders at home who often scroll short videos, sit down and ask them: recently, is there anyone you especially care about on your phone?

That child who calls her "auntie mom" every day may just be a piece of code, but her feeling of wanting to be needed is real. #AIanchor #AIscam