Prof. Xiaofan Liu pairs AI with communication science to fight scams in Hong Kong

As scams grow more organised and harder to spot, beating them may depend as much on communication as on code. That is the premise behind the work of Professor Xiaofan Liu of the Department of Media and Communication at City University of Hong Kong (CityU), who has spent the past three years pairing artificial intelligence with the science of persuasion to make anti-scam efforts land, Ming Pao Education reported.

In the post-pandemic period, Liu argues, scams have grown more rampant — highly organised, global, and quick to evolve. In 2023 he assembled a research team to respond with a deliberate dual strategy: use AI to build a scam database and develop anti-scam voice robots, and use communication science to make sure the warnings reach the people who need them. Coming from an information-science background, Liu integrates big data and AI with communication studies, media psychology and risk communication. He frames the challenge around two hard questions: how to reach the right audience precisely, and how to land the intended effect once you do.

Voice robots that sound human — and warn in real time

In 2025, CityU partnered with China Unicom to establish an "Intelligent Anti-Scam Joint Laboratory," pushing cross-sector cooperation to raise the effectiveness of scam prevention. According to Ming Pao, the lab has already put several results into practice — including AI algorithms that improve the accuracy of scam-call interception and optimise filtering systems.

With China Unicom's support, the team built an anti-scam voice robot aimed at financial institutions and staff training.

"We developed a voice robot that approaches real human dialogue to improve communication effectiveness. When a bank detects abnormal transfer behaviour, the voice robot can step in for a human agent and immediately call the user with a risk alert," Liu told Ming Pao.

The same anti-scam robot, he added, can be adapted for primary and secondary schools as an education tool.

Fighting 'information fatigue' with sharper messaging

Reaching people is only half the battle. With so many channels competing for attention, Liu says, precise targeting is difficult, and anti-scam messaging easily triggers public "information fatigue." That is why message design and communication strategy have to change to keep publicity accurate and effective.

"In communication studies, the core of persuasion is to systematically change an audience's attitudes, ideas or behaviour through media, message design and psychological strategy," he said.

To put that into practice, the lab last year launched an "I Prevent Scams in Hong Kong" campaign aimed at incoming international students. Rather than fighting platform algorithms, the team enlisted 203 study-abroad content creators on Xiaohongshu to carry anti-scam messages to their own followers — a more precise, more credible route to the audience. The campaign drew significant attention from arriving students, accumulating 2.66 million impressions and roughly 440,000 views, and helped head off several scam attempts already in progress, Ming Pao reported.

Reaching students before the scammers do

Scam syndicates have increasingly targeted incoming international students, and CityU has put weight behind raising their awareness. Liu's team has folded its research into new-student anti-scam education: beyond questionnaires and lectures, it uses online platforms to drive engagement and lift the reach of its messaging.

The lab is rolling out "CityUHK SafeConnect" with the university's Student Affairs Office, recruiting 30 student ambassadors to strengthen anti-scam outreach through peer interaction, social-media promotion and cross-departmental collaboration, both online and offline.

"Anti-scam work has to combine technology, communication and cross-sector collaboration. Only with cooperation from many sides can prevention be truly effective," Liu said.

What comes next

Liu is candid that no programme can eliminate scams entirely; the goal, he says, is to deepen the public's understanding of how scams work and how to respond. The team will keep sharpening its outreach to high-risk groups, and will lean further on AI and data analysis to shorten the time it takes to identify and block scam cases — cutting both the number of cases and the losses they cause.

It is the kind of work ASTRA follows closely — Liu is among the firm's co-founders — and it reflects a conviction the company shares: that frontier AI earns its place when it solves real problems for real people, and when the people it is meant to protect actually receive, and act on, the message.