IAPQB

The finance chief on the video call was fake

Deepfake payment fraud in Hong Kong and Singapore, and the checks police and regulators recommend

IAPQB Report No. 28 · Case file · Published 17 November 2026
International AI Practice Qualification Board (IAPQB)

In one minute

  • In January 2024, a Hong Kong finance employee of Arup authorised transfers of about HK$200 million (US$25.6 million) after a video meeting with what looked like the UK finance chief and colleagues. Police believed the meeting was pre-recorded [1, 2, 3].
  • Hong Kong had a second case in May 2024, and Singapore police counted at least 10 cases of fake executives on WhatsApp from January 2025 to February 2026, with losses of at least S$13.5 million [1, 4].
  • In April 2024, a LastPass employee ignored a cloned voice of its chief executive on WhatsApp because it came outside normal business channels and showed forced urgency [5].
  • Police and regulators recommend checks on a second, known channel, a second approver, code words agreed in advance, briefed staff and fast reports to the bank [6, 7, 8, 9].

1. A familiar situation

A finance officer gets a message from the group's finance chief abroad about a confidential deal. It looks like phishing. Then, in a video meeting, the finance chief appears with colleagues he knows by sight. In the Hong Kong case below, such a meeting removed the employee's doubts; every other person in it was fake [1, 2].

2. Why it works

Payment approvals often rest on recognising a senior person, and deepfake tools copy faces and voices from public recordings. Hong Kong police believed the January 2024 meeting was pre-recorded from public video clips and voices, with no real exchange: the fraudster gave the instruction, ended the meeting with an excuse and continued by instant message [1].

The scripts also block checks. The Hong Kong message spoke of a secret transaction [2], and Singapore victims are made to sign non-disclosure agreements that keep them from asking colleagues [4]. The FBI says AI-generated content "is often difficult to identify" [10].

IAPQB view. In Hong Kong and Singapore, the staff who paid had seen what looked like a senior executive and others in a video meeting, and the meeting was part of the fraud [1, 11]. Where a loss was avoided or recovered, the helpful step came outside the fraudsters' channel: a report to LastPass's security team, and an alert to the bank in Singapore [5, 11].

3. Four cases

3.1 Hong Kong, 2024: every other face in the meeting was fake

In late January 2024, an employee of a multinational firm told Hong Kong police that a fake chief financial officer of the UK head office had invited him to a group video conference about confidential transactions. He authorised transfers to five local bank accounts, losing about HK$200 million [1]. He first suspected phishing, police said, but the people on the call looked and sounded like colleagues he knew. He found out when he checked with head office [2].

In May 2024, Arup confirmed to CNN that it was the target, that it had told police in January and that "fake voices and images were used"; it said no internal systems were compromised [3]. In June 2024 the case was still under investigation [1]. Arup's chief information officer later made a real-time deepfake of himself with open-source software in "about 45 minutes", though not a particularly convincing one [12].

3.2 LastPass, 2024: a cloned chief executive on WhatsApp

In April 2024, LastPass reported that an employee had received calls, texts and at least one voicemail on WhatsApp with an audio deepfake of its chief executive. The contact came outside normal business channels and showed signs such as forced urgency, so the employee ignored it and told the internal security team. LastPass said there was no impact on the company [5].

3.3 Singapore, 2025: a fake Zoom call, and money held in Hong Kong

On 24 March 2025, a scammer posing as a multinational company's chief financial officer contacted its finance director on WhatsApp. A fake lawyer stressed confidentiality, and the director signed a non-disclosure agreement. On 25 March he joined a Zoom call in which the chief executive and others were impersonated with deepfakes. The next day he sent over US$499,000 to a local account controlled by the scammers, and over US$494,000 moved on to Hong Kong. Asked for another US$1.4 million on 27 March, he realised it was a scam and alerted the company's bank, HSBC, which told police. On 28 March, Hong Kong police withheld the full amount there. Investigations were ongoing in April 2025 [11].

3.4 Hong Kong and Singapore, 2024 to 2026: the script returns

On 20 May 2024, a fake UK-based chief financial officer messaged an employee of a multinational trade company in Hong Kong on WhatsApp, then, in a nearly 30-minute video conference, told the employee to transfer nearly HK$4 million [1]. Hong Kong police received three deepfake fraud cases in 2024, including impersonation of senior company executives [13]. From January 2025 to February 2026, Singapore police counted at least 10 cases of executives impersonated on WhatsApp, some with digitally altered Zoom calls, and losses of at least S$13.5 million [4].

4. How practitioners handle it now

Requests are checked on a second, known channel. The FBI advises hanging up and calling back on a number found independently [14]; its 2017 advice was to use "previously known numbers" [6]. The UK National Cyber Security Centre asks that important email requests be verified "using a second type of communication" [7]. For video calls, the Monetary Authority of Singapore (MAS) suggests a different medium, such as an in-person conversation [8].

A second person approves. MAS's September 2025 information paper on deepfakes says institutions should split critical duties to reduce the chance that one deceived person can execute transactions for the whole organisation, with "dual control", where two people act on a task together [8]. The FBI's 2017 advice included "a secondary sign-off by company personnel" for changed supplier payment details [6].

Code words are agreed in advance. MAS suggests "code words", pre-agreed secret phrases changed periodically, for requests made through remote channels [8]. In 2018 the FBI suggested code phrases "that would only be known to the two legitimate parties" [15]. Hong Kong police suggest asking a person on a video call to turn their head, or testing them on the conversation [1].

Staff who move money are briefed and practise. A joint advisory from Singapore's police, MAS and Cyber Security Agency asks businesses to alert employees "responsible for making fund transfers" [9], and police later added new employees and interns [4]. MAS suggests deepfake simulation exercises, and US agencies recommend tabletop exercises that include executives [8, 16].

Fraud is reported to the bank at once. The FBI says "time is of the essence" and the bank should be asked to recall the funds [17]. In Singapore, the finance director quickly alerted HSBC, which immediately told police, and the money was held the day after he realised [11]. Since November 2024, the US Treasury's Financial Crimes Enforcement Network has asked financial institutions to tag reports of suspected deepfake fraud "FIN-2024-DEEPFAKEFRAUD" [18].

5. What to take from this

For staff who approve or send payments. In Hong Kong, the people on screen looked and sounded like colleagues [2], and in both cities the call was part of the fraud [1, 11]. Most checks described by police and regulators do not rely on what the call shows: a call back on a number already on file, a code word agreed beforehand, or a word face to face [6, 7, 8].

For finance managers and boards. The cases in 3.1 and 3.3 came with a request for secrecy [2, 11]. A written rule that confidentiality never cancels the second check is aimed at that step, and dual control adds a second person to the payment itself [8].

For schools, training providers and employers. Police and regulators ask businesses to train and brief staff, especially those who make fund transfers, new employees and interns [4, 9]. Exercises built on these cases can rehearse the real scripts: a secret deal, a non-disclosure agreement and a larger second request [2, 8, 11].

6. Limits of this report

Hong Kong authorities did not name Arup; the name comes from Arup's own statements [3, 12]. No official update on that case after June 2024 was found [1]. Singapore's figures are cases reported to police [4]. A 2019 case whose main reports could not be opened is left out: a voice-clone theft of US$243,000 from a UK company that US agencies mention in one line [16]. The FBI advice on code phrases and sign-offs dates from 2017 and 2018, and the MAS paper lists measures institutions could adopt [6, 8, 15]. This report tests no detection tool.

References

[1] HKSAR Government (26 June 2024). LCQ9: Combating frauds involving deepfake [written reply by the Secretary for Security, Mr Tang Ping-keung, to a question by Dr the Hon Tan Yueheng, Legislative Council]. Press release. https://www.info.gov.hk/gia/general/202406/26/P2024062600192.htm

[2] Chen, H., & Magramo, K. (4 February 2024). Finance worker pays out $25 million after video call with deepfake ‘chief financial officer’. CNN. https://edition.cnn.com/2024/02/04/asia/deepfake-cfo-scam-hong-kong-intl-hnk/index.html

[3] Magramo, K. (16 May 2024; updated 17 May 2024). British engineering giant Arup revealed as $25 million deepfake scam victim. CNN Business. https://edition.cnn.com/2024/05/16/tech/arup-deepfake-scam-loss-hong-kong-intl-hnk/index.html

[4] Singapore Police Force (26 February 2026). Police Advisory On Scams Involving Impersonation Of Companies’ Senior Executives On WhatAapp [spelling as published]. Police news release. https://www.police.gov.sg/media-hub/news/2026/02/20260226_police_advisory_on_scams_involving_impersonation_of_companies_on_whatsapp

[5] Kosak, M. (10 April 2024). Attempted Audio Deepfake Call Targets LastPass Employee. LastPass blog. https://blog.lastpass.com/posts/attempted-audio-deepfake-call-targets-lastpass-employee

[6] Federal Bureau of Investigation, Internet Crime Complaint Center (4 May 2017). Business E-mail Compromise E-mail Account Compromise The 5 Billion Dollar Scam. Public service announcement I-050417-PSA. https://www.ic3.gov/PSA/2017/PSA170504

[7] National Cyber Security Centre (UK) (5 February 2018; reviewed 13 February 2024). Phishing attacks: defending your organisation. https://www.ncsc.gov.uk/guidance/phishing

[8] Monetary Authority of Singapore (September 2025). Cyber Risks Associated with Deepfakes. Information paper, published with circular MAS/TCRS/2025/06 on 18 September 2025. https://www.mas.gov.sg/-/media/mas-media-library/regulation/circulars/trpd/cyber-risks-associated-with-deepfakes.pdf (circular page: https://www.mas.gov.sg/regulation/circulars/cyber-risks-associated-with-deepfakes)

[9] Singapore Police Force, Monetary Authority of Singapore and Cyber Security Agency of Singapore (12 March 2025). Joint Advisory On Scams Involving Digital Manipulation. Police news release. https://www.police.gov.sg/media-hub/news/2025/03/20250312_joint_advisory_on_scams_involving_digital_manipulation (also published by the Cyber Security Agency of Singapore: https://www.csa.gov.sg/alerts-and-advisories/advisories/ad-2025-005/)

[10] Federal Bureau of Investigation, Internet Crime Complaint Center (15 May 2025). Senior US Officials Impersonated in Malicious Messaging Campaign. Public service announcement I-051525-PSA. https://www.ic3.gov/PSA/2025/PSA250515

[11] Singapore Police Force (7 April 2025). Singapore And Hong Kong Police Force Recover Over $670,000 In A Scam Involving Digital Impersonation. Police news release. https://www.police.gov.sg/media-hub/news/2025/04/20250407_singapore_and_hong_kong_police_force_recover_over_-670000_in_a_scam

[12] Elliott, D. (4 February 2025). ‘This happens more frequently than people realize’: Arup chief on the lessons learned from a $25m deepfake crime. World Economic Forum, Forum Stories. https://www.weforum.org/stories/2025/02/deepfake-ai-cybercrime-arup/

[13] HKSAR Government (26 February 2025). LCQ11: Prevention of telephone fraud [written reply by the Secretary for Commerce and Economic Development, Mr Algernon Yau, to a question by Dr the Hon Dennis Lam, Legislative Council]. Press release. https://www.info.gov.hk/gia/general/202502/26/P2025022600236.htm

[14] Federal Bureau of Investigation, Internet Crime Complaint Center (3 December 2024). Criminals Use Generative Artificial Intelligence to Facilitate Financial Fraud. Public service announcement I-120324-PSA. https://www.ic3.gov/PSA/2024/PSA241203

[15] Federal Bureau of Investigation, Internet Crime Complaint Center (12 July 2018). Business E-mail Compromise The 12 Billion Dollar Scam. Public service announcement I-071218-PSA. https://www.ic3.gov/PSA/2018/PSA180712

[16] National Security Agency, Federal Bureau of Investigation and Cybersecurity and Infrastructure Security Agency (September 2023). Contextualizing Deepfake Threats to Organizations. Cybersecurity Information Sheet, version 1.0. https://media.defense.gov/2023/Sep/12/2003298925/-1/-1/0/CSI-DEEPFAKE-THREATS.PDF

[17] Federal Bureau of Investigation, Internet Crime Complaint Center (11 September 2024). Business Email Compromise: The $55 Billion Scam. Public service announcement I-091124-PSA. https://www.ic3.gov/PSA/2024/PSA240911

[18] Financial Crimes Enforcement Network, US Department of the Treasury (13 November 2024). FinCEN Alert on Fraud Schemes Involving Deepfake Media Targeting Financial Institutions. FIN-2024-Alert004. https://www.fincen.gov/system/files/shared/FinCEN-Alert-DeepFakes-Alert508FINAL.pdf

How to cite

IAPQB (2026). The finance chief on the video call was fake: deepfake payment fraud in Hong Kong and Singapore, and the checks police and regulators recommend. IAPQB Report No. 28. International AI Practice Qualification Board. https://iapqb.org/research/028-deepfake-video-call-fraud.html

About this report

Prepared by IAPQB. Each fact comes from the sources listed, and each source was opened and checked on 16 November 2026; anything that could not be checked was left out. AI tools were used to search and to draft, and IAPQB checked every claim against its source and is responsible for the text. IAPQB runs an AI work qualification; this report does not promote it. Corrections: contact@iapqb.org or an issue at github.com/iapqb/iapqb.github.io. Licence: CC BY 4.0.

Archived with the series on Zenodo: doi.org/10.5281/zenodo.24248199 · Markdown sources: github.com/iapqb/research