In early 2024, a Hong Kong finance employee was targeted in what is today described as one of the most sophisticated corporate AI scams ever committed in the city. Fraudsters used AI deepfakes to trick an employee of a multinational engineering firm into transferring HK$200 million.
Residents of Hong Kong lost an estimated US$104.8 million to deepfake-enabled scams, with almost half the losses traced to social media, according to a new study by Surfshark.
The analysis places Hong Kong eighth globally among 48 countries surveyed and highlights how artificial-media fraud has moved from novelty to a mainstream threat across routine digital channels.
Social platforms drive majority of losses
Surfshark’s research shows social media accounted for US$47.1 million, or 44.9% of the city’s deepfake-related losses. Phone calls were the second-largest source at roughly US$35 million (33.4%).
Globally, the study estimates at least US$3.7 billion has been lost to deepfake scams to date, with social networks responsible for about 47% of that total — roughly US$1.73 billion.
The report underlines how a single forged clip or convincing message can reach millions in hours, creating fertile ground for investment fraud, impersonation and other financially motivated deceptions.
Surfshark’s research lead, Luís Costa, warned that “social media platforms give scammers instant access to billions of users simultaneously. A single deepfake video can go viral in hours, reaching millions of potential victims before it's detected or removed.”
Impersonation and routine channels rising
Impersonation fraud — where criminals use deepfake techniques to mimic individuals and bypass identity checks — represented the second-largest category globally, accounting for US$911 million or about 25% of losses.
Other notable origins include fake job-candidate schemes (around US$100 million) and conventional communication channels: phone calls (US$71 million), video platforms (US$62 million) and messaging apps (US$41 million), which together added roughly US$174 million in losses.
Costa said the findings show an expanding threat landscape: “Although social media is the most prominent source of deepfake fraud, the growing volume of attacks, even via regular phone calls or messaging apps, shows a dangerous expansion into routine interactions. To protect ourselves, we must shift our mindset: awareness is the only effective defence when a familiar voice or face can no longer be trusted implicitly.”
Practical defences for individuals and families
The Surfshark report includes pragmatic steps to reduce risk. It recommends scepticism towards unsolicited investment claims on social channels and verification through official sources before transferring funds.
Households are advised to set a family “safe word” to confirm identity during urgent calls, and to perform simple “glitch tests” in video calls — such as asking the caller to wave — because deepfakes often distort when facial features are obscured.
Other measures include restricting high-resolution media online and strengthening multi-factor authentication with hardware tokens or biometrics rather than SMS codes.
As synthetic media tools become more accessible, the study suggests businesses, regulators and the public must combine technical controls with heightened user awareness to blunt the rising tide of financially motivated deepfakes.
