A police insider said the higher losses could be partly attributed to two separate deepfake scams that were still under investigation.

Criminals used artificial intelligence (AI) to alter voices and faces and fooled employees at two UK-based multinational firms in the city into transferring HK$204 million to them.

London-based design and engineering firm Arup lost HK$200 million after one of its Hong Kong employees was taken in by a digitally recreated version of the firm’s chief financial officer in January.

Police first noticed deepfakes last year when they arrested nine people in a crackdown on a fraud syndicate that used AI technology to doctor images for loan scams targeting moneylenders. The investigation involved HK$200,000 in stolen funds.

Secretary for Security Chris Tang Ping-keung earlier revealed that police had discovered 21 online video clips from last November to May where AI technology was used to impersonate government officials or celebrities to promote questionable investment products.

He said the videos had been removed from online platforms and no reports had been received of the public being defrauded directly as a result of the deepfake videos.

The scams that involved the greatest losses from January to May were sham investment deals, which duped 1,503 people out of HK$915 million.

There were 1,243 victims tricked out of HK$624 million in the same period last year.

The source said this year’s figures highlighted the continued threat from investment scams that preyed on Hongkongers’ hopes of making easy money.

He warned the public to be wary of “unrealistic promises of high returns with low risks”.

“Scammers are still sending text messages to the public and inviting them to join VIP chat groups, where they impersonate investment experts or fund managers to deceive victims into purchasing shares or cryptocurrency on bogus trading platforms,” he said.

The source added the ruse had been around for several years and that con artists had also impersonated celebrities and influencers to cheat people.

The force warned on its website that, in another type of tactic known as “pig-butchering scams”, “scammers establish an online romantic relationship to gain the trust of the victims then lure them into making an investment”.

“Beware of the investment projects recommended by the perfect men or women you meet online,” it said.

Police figures showed that internet shopping scam losses jumped by 73 per cent to HK$133 million despite an increase of just seven cases to 4,175 in the first five months of the year.

But the amounts lost in employment, romance and email scams dropped.

Although May recorded the highest number of online employment scams in a month, at 348, the overall number of such fraud cases fell by 38 per cent to 1,118. Losses from these scams also fell by 31 per cent to HK$252 million.

The number of email scams increased by 39 per cent to 99, but losses dropped by 24 per cent to HK$38 million.

There were 409 reports of online romance scams between January and May this year, down by 28 per cent from the 574 cases logged in the same period last year.

Criminals pocketed HK$204 million from such frauds, down by 35 per cent from HK$315 million the year before.

A police spokesman said that the force “works closely with local and overseas law enforcement agencies and actively mounts operations based on intelligence gathered online to minimise the threat posed by internet deception”.

He added police also collaborated with the Hong Kong Monetary Authority, banks and stored value facility providers to boost payment mechanism security and alert users to potential loopholes.

The spokesman said police also engaged with the Office of the Communications Authority, mobile network operators and online service providers to exchange information on cyber criminals’ activities, to take down criminal content or suspend scamming accounts from the internet.

Police appealed to the public to use the force’s Scameter search engine, accessible through the CyberDefender website or app, to check for suspicious or fraudulent schemes.

The search engine has information that can help people identify suspicious web addresses, emails, platform usernames, bank accounts, mobile phone numbers and IP addresses.