Detection queries Reported campaignBusiness & social

Arup Hong Kong deepfake video-call fraud (~US$25.6M)

A finance employee on a video call with deepfaked executives approved transfers of about US$25.6M. Arup said its internal IT systems were not compromised.

What happened

Jan 2024: Hong Kong finance employee joined video call with deepfake of UK CFO and colleagues; authorized 15 transfers totaling ~HK$200M (~US$25.6M) to five local accounts. Arup said internal IT systems were not compromised.

Deepfake-enabled BEC/fraud (video call impersonation); internal systems not compromised

Read the original sourceAdditional source: theguardian.com

How it happened

Select a stage to explore its defenses
Detect this stage
Detect this stage
Detect this stage
Select a stage · Drag to explore · + / − to zoom · Dashed edges are loopsSelect a stage · Arrows show reported connectionsReported sequence
Read the flow as text & sources

A catalog reconstruction of the supplied source account. Unknown steps remain explicit; an edge describes the reported relationship between steps.

  1. 01 · Finance employee joins a video call

    The victim is approached through a meeting impersonating company executives.

    Hunt: A meeting or message requests a new beneficiary, split transfers or an override of normal payment approval.

    A video call's apparent identity is not authentication. Endpoint application control and malware hunts do not detect a pure business-process fraud.

    Source for step 1
  2. 02 · Deepfake executives appear on screen

    Synthetic video and audio impersonate the UK CFO and colleagues.

    Source for step 2
  3. 03 · Payment requests are authorized

    The employee approves 15 transfers to five local accounts.

    Hunt: A meeting or message requests a new beneficiary, split transfers or an override of normal payment approval.

    A video call's apparent identity is not authentication. Endpoint application control and malware hunts do not detect a pure business-process fraud.

    Source for step 3
  4. 04 · Approximately US$25.6M lost

    Arup disclosed the loss and said its internal IT systems were not compromised.

    Hunt: A meeting or message requests a new beneficiary, split transfers or an override of normal payment approval.

    A video call's apparent identity is not authentication. Endpoint application control and malware hunts do not detect a pure business-process fraud.

    Source for step 4

Why it matters

Reported entry point: Social engineering via deepfake video conference. Reported impact: Arup disclosed the loss and said its internal IT systems were not compromised; reported targets include Arup (engineering consultancy) Hong Kong finance staff; applicability to your setup has not been assessed, so verify affected versions, permissions, and configuration in the original report.

Read the original report

What you can do

Protection & prevention

Out-of-band payment verification; dual control; challenge deepfake meetings

Catalog defensive guidance based on the reported behavior. Confirm scope and current mitigations in the original source.

Review the original guidance Additional source 2
Investigate each phase

Detection queries

Draft query starters grouped by the attack stages where they apply.

Download stage pack

Start with the logs you collect. Each hunt identifies its required visibility, field mapping and limits. Queries are catalog-authored drafts using reusable behavior profiles; test with known activity and your normal baseline before alerting.

01Finance employee joins a video callOut-of-band payment verificationCustom schema · Draft

The victim is approached through a meeting impersonating company executives.

What to look for

A meeting or message requests a new beneficiary, split transfers or an override of normal payment approval.

Visibility: Defender-side logs where collected

Required logs

  • Treasury / payment approval and beneficiary-change audit
  • Meeting and communication metadata
  • Independent callback / dual-approval records

Coverage limit. A video call's apparent identity is not authentication. Endpoint application control and malware hunts do not detect a pure business-process fraud.

Custom-schema candidates, not out-of-box detectors. Map your application / cloud / platform logs to EventType, Actor, Asset, Action, Target and CorrelationId. KQL uses a custom OpenAttacksAudit_CL table; SPL uses openattacks:audit. EventType values are proposed normalization names, not native vendor event names. Correlation and environment-specific alert logic remain to be built.

Sigma
Draft SigmaDownload Sigma
title: Hunt - Out-of-band payment verification
id: 813e417f-8751-4479-ad17-0f3d146ff036
status: experimental
description: A meeting or message requests a new beneficiary, split transfers or
  an override of normal payment approval.
author: MagicSword OpenAttacks catalog
date: 2026-10-23
falsepositives:
  - Authorized activity matching this broad hunt; baseline and investigate
    before alerting.
level: low
logsource:
  category: application
  product: openattacks_audit
  definition: Custom normalized audit schema; map EventType, Actor, Asset, Action,
    Target and CorrelationId from the required source logs first.
detection:
  selection:
    EventType:
      - payment.beneficiary.change
      - payment.approval.override
      - payment.transfer
  condition: selection
KQL
Draft KQLDownload KQL
// SCHEMA TEMPLATE: create/map this custom table before use; not a built-in Sentinel table.
// OpenAttacksAudit_CL: TimeGenerated(datetime), EventType/Actor/Asset/Action/Target/CorrelationId(string).
// Selects candidates only. Apply the stage hypothesis and correlation steps before alerting.
OpenAttacksAudit_CL
| where TimeGenerated > ago(7d)
| where EventType in ("payment.beneficiary.change", "payment.approval.override", "payment.transfer")
| project TimeGenerated, EventType, Actor, Asset, Action, Target, CorrelationId
| order by TimeGenerated desc
SPL
Draft SPLDownload SPL
sourcetype="openattacks:audit"
EventType IN ("payment.beneficiary.change", "payment.approval.override", "payment.transfer")
| table _time EventType Actor Asset Action Target CorrelationId
| sort - _time
Investigation steps
  1. Review new payees, transfer bursts and approval overrides against established payment policy.
  2. Verify the request using a known contact channel; preserve the meeting record and authorization history.
03Payment requests are authorizedOut-of-band payment verificationCustom schema · Draft

The employee approves 15 transfers to five local accounts.

What to look for

A meeting or message requests a new beneficiary, split transfers or an override of normal payment approval.

Visibility: Defender-side logs where collected

Required logs

  • Treasury / payment approval and beneficiary-change audit
  • Meeting and communication metadata
  • Independent callback / dual-approval records

Coverage limit. A video call's apparent identity is not authentication. Endpoint application control and malware hunts do not detect a pure business-process fraud.

Custom-schema candidates, not out-of-box detectors. Map your application / cloud / platform logs to EventType, Actor, Asset, Action, Target and CorrelationId. KQL uses a custom OpenAttacksAudit_CL table; SPL uses openattacks:audit. EventType values are proposed normalization names, not native vendor event names. Correlation and environment-specific alert logic remain to be built.

Sigma
Draft SigmaDownload Sigma
title: Hunt - Out-of-band payment verification
id: 813e417f-8751-4479-ad17-0f3d146ff036
status: experimental
description: A meeting or message requests a new beneficiary, split transfers or
  an override of normal payment approval.
author: MagicSword OpenAttacks catalog
date: 2026-10-23
falsepositives:
  - Authorized activity matching this broad hunt; baseline and investigate
    before alerting.
level: low
logsource:
  category: application
  product: openattacks_audit
  definition: Custom normalized audit schema; map EventType, Actor, Asset, Action,
    Target and CorrelationId from the required source logs first.
detection:
  selection:
    EventType:
      - payment.beneficiary.change
      - payment.approval.override
      - payment.transfer
  condition: selection
KQL
Draft KQLDownload KQL
// SCHEMA TEMPLATE: create/map this custom table before use; not a built-in Sentinel table.
// OpenAttacksAudit_CL: TimeGenerated(datetime), EventType/Actor/Asset/Action/Target/CorrelationId(string).
// Selects candidates only. Apply the stage hypothesis and correlation steps before alerting.
OpenAttacksAudit_CL
| where TimeGenerated > ago(7d)
| where EventType in ("payment.beneficiary.change", "payment.approval.override", "payment.transfer")
| project TimeGenerated, EventType, Actor, Asset, Action, Target, CorrelationId
| order by TimeGenerated desc
SPL
Draft SPLDownload SPL
sourcetype="openattacks:audit"
EventType IN ("payment.beneficiary.change", "payment.approval.override", "payment.transfer")
| table _time EventType Actor Asset Action Target CorrelationId
| sort - _time
Investigation steps
  1. Review new payees, transfer bursts and approval overrides against established payment policy.
  2. Verify the request using a known contact channel; preserve the meeting record and authorization history.
04Approximately US$25.6M lostOut-of-band payment verificationCustom schema · Draft

Arup disclosed the loss and said its internal IT systems were not compromised.

What to look for

A meeting or message requests a new beneficiary, split transfers or an override of normal payment approval.

Payment and approval records can support investigation of the loss; they do not prove that media was synthetic.

Visibility: Defender-side logs where collected

Required logs

  • Treasury / payment approval and beneficiary-change audit
  • Meeting and communication metadata
  • Independent callback / dual-approval records

Coverage limit. A video call's apparent identity is not authentication. Endpoint application control and malware hunts do not detect a pure business-process fraud.

Custom-schema candidates, not out-of-box detectors. Map your application / cloud / platform logs to EventType, Actor, Asset, Action, Target and CorrelationId. KQL uses a custom OpenAttacksAudit_CL table; SPL uses openattacks:audit. EventType values are proposed normalization names, not native vendor event names. Correlation and environment-specific alert logic remain to be built.

Sigma
Draft SigmaDownload Sigma
title: Hunt - Out-of-band payment verification
id: 813e417f-8751-4479-ad17-0f3d146ff036
status: experimental
description: A meeting or message requests a new beneficiary, split transfers or
  an override of normal payment approval.
author: MagicSword OpenAttacks catalog
date: 2026-10-23
falsepositives:
  - Authorized activity matching this broad hunt; baseline and investigate
    before alerting.
level: low
logsource:
  category: application
  product: openattacks_audit
  definition: Custom normalized audit schema; map EventType, Actor, Asset, Action,
    Target and CorrelationId from the required source logs first.
detection:
  selection:
    EventType:
      - payment.beneficiary.change
      - payment.approval.override
      - payment.transfer
  condition: selection
KQL
Draft KQLDownload KQL
// SCHEMA TEMPLATE: create/map this custom table before use; not a built-in Sentinel table.
// OpenAttacksAudit_CL: TimeGenerated(datetime), EventType/Actor/Asset/Action/Target/CorrelationId(string).
// Selects candidates only. Apply the stage hypothesis and correlation steps before alerting.
OpenAttacksAudit_CL
| where TimeGenerated > ago(7d)
| where EventType in ("payment.beneficiary.change", "payment.approval.override", "payment.transfer")
| project TimeGenerated, EventType, Actor, Asset, Action, Target, CorrelationId
| order by TimeGenerated desc
SPL
Draft SPLDownload SPL
sourcetype="openattacks:audit"
EventType IN ("payment.beneficiary.change", "payment.approval.override", "payment.transfer")
| table _time EventType Actor Asset Action Target CorrelationId
| sort - _time
Investigation steps
  1. Review new payees, transfer bursts and approval overrides against established payment policy.
  2. Verify the request using a known contact channel; preserve the meeting record and authorization history.
Technical detailsTooling, impact & MITRE ATLAS

Attack context

AI's role
Other AI-enabled attacks
Initial access
Social engineering via deepfake video conference

Tooling as reported

AI products, models and artifacts

  • Deepfake video/audio of executives

Targets and reported impact

Targets / sectors
Arup (engineering consultancy) Hong Kong finance staff
Reported scale / exposure
~US$25.6 million loss
Target geography
Hong Kong

The affected finance staff were in Hong Kong. The United Kingdom refers to the parent organization and executive impersonation, not a separately confirmed victim location.

Detection rules

  • Review attack-specific indicatorsGUIDANCE

    Monitor for the techniques described in Arup Hong Kong deepfake video-conference $25M fraud. Refer to sources for specific IOCs and TTPs.

Source defensive notes

Out-of-band payment verification; dual control; challenge deepfake meetings

MITRE ATLAS

AI-specific catalog annotations based on the reported behavior; these are not MITRE case-study assignments.

AML.T0088Generate Deepfakes

Synthetic video and audio impersonate executives. Source

AML.T0073Impersonation

The payment request relies on executive impersonation. Source

Sources & research credit

2 sources

Research by Arup / Hong Kong Police (press) · Arup (victim disclosure via press)

  1. cnn.com
  2. theguardian.com
Record & downloads
Source record name
Arup Hong Kong deepfake video-conference $25M fraud
Entry ID
arup-deepfake-video-fraud-2024-01
Catalog review
October 20, 2026
Category
adversary-use-of-ai
AI involvement confidence
confirmed