Rising component costs may adjust prices on selected items without prior notice. Supply constraints may add shipping time — we’ll notify you if affected. OEM costs and supply tightening — contact us to secure stock and pricing. For projects or bulk orders, choose “Request for Quotation” on any item.
Call our specialists(65) 6978 3502 Email our sales teamsales@sourceit.com.sg

Networking & Cybersecurity

Could Deepfake Voices Become the Next Corporate Conference Call Scam?

Deepfake voices are already being used in corporate conference call scams. This isn't a future problem. Reputable outlets have reported cases where cloned voices and faces of senior executives were used in video meetings to push staff into transferring money. Singapore's police, central bank and cyber security agency issued a joint advisory on this tactic in March 2025. The best defence is a verification process that doesn't depend on recognising a voice or face.

This post covers what has actually been reported, how these scams tend to unfold, why familiar voices are no longer proof of identity, and the controls Singapore businesses can put in place. For the wider picture of AI-related security exposure, see Your AI Assistant Might Be Your Biggest Security Risk.

Has this actually happened to real companies?

Yes. Two widely reported cases show both outcomes.

Arup, Hong Kong (2024). CNN reported that a finance employee in the Hong Kong office of engineering firm Arup transferred HK$200 million, about US$25.6 million, across 15 transactions. According to Hong Kong police, as reported by CNN, the employee was first suspicious of an email asking for a confidential transaction. He set those doubts aside after a video call with people who looked and sounded like the chief financial officer and other colleagues, all of them deepfake recreations. Arup told CNN, "we can confirm that fake voices and images were used".

WPP (2024). The Guardian reported an attempt targeting WPP, the advertising group. Scammers set up a WhatsApp account with a public photo of chief executive Mark Read and used it to arrange a Microsoft Teams meeting. In the meeting they used an AI voice clone of a senior executive along with YouTube footage, and posed as Read in the chat. They asked an agency leader to set up a new business in order to extract money and personal details. The attempt failed, and WPP credited the vigilance of its staff.

What have Singapore authorities warned about?

On 12 March 2025 the Singapore Police Force, the Monetary Authority of Singapore and the Cyber Security Agency of Singapore issued a joint advisory on scams involving digital manipulation. It describes this pattern:

  • Employees receive unsolicited WhatsApp messages from people claiming to be their company's executives, inviting them to a Zoom call.
  • On the call, digital manipulation is believed to have been used to make the scammers look like the real executives. Some calls also involve people posing as MAS officials or investors.
  • Victims are told to transfer substantial funds from corporate accounts, framed as project financing or investments, and are sometimes asked for NRIC or passport details.
  • A second scammer posing as legal counsel may send documents such as a non-disclosure agreement to the victim's personal email.

The advisory urges businesses to set up protocols so employees can verify the authenticity of video calls and messages, especially those that appear to come from senior executives. It notes that MAS will never ask anyone to transfer money.

CSA's earlier advisory on detecting and responding to deepfake scams (March 2024) lists warning signs. These include lips out of sync with speech, unnatural blinking, blurred edges around the face and limited variation in tone. CSA also cautions that consumer deepfake detection tools are still immature.

Why is a familiar voice no longer proof of identity?

Senior executives are easy to clone. Many of them speak at conferences, on earnings calls and in company videos, which gives scammers plenty of source audio. The WPP attempt reportedly used YouTube footage of the executives involved.

The US FBI made the same point in a May 2025 public service announcement about AI-generated voice messages impersonating senior US officials. Its advice: "do not assume it is authentic" when a message claims to come from a senior official. The FBI suggests listening closely to tone and word choice, but it also tells people to verify identity through a separately confirmed number. Checking a voice by ear alone isn't enough.

Scammers also use pressure. Both reported cases and the Singapore advisory share the same features: urgency, secrecy, a senior figure and an unusual payment. Spotting the pattern is more reliable than spotting glitches in the audio.

Which controls actually stop a deepfake call?

The controls that work don't depend on how convincing the fake is. They change what a single call is allowed to authorise.

Control How it helps Who owns it
Call-back on a known number Hang up and call the executive on a number from the company directory, not one supplied in the message Finance, executive assistants
No payment authority from calls or chats alone A video call, voice note or WhatsApp message can never be the sole approval for a transfer or change of bank details CFO, finance controller
Dual approval above a threshold A second person must independently approve payments over a set amount, so one fooled employee cannot release funds Finance
Challenge question or code phrase Ask something only the real person would know; the FBI suggests an agreed secret word Executives and their teams
Official channels only Executives agree never to run payment requests through personal WhatsApp or unscheduled meetings Leadership
External access limits in Teams Restrict which outside domains can chat or meet with staff so impostor accounts stand out IT
Staff awareness Brief anyone who handles payments on the reported patterns and the CSA warning signs HR, IT security

On the Microsoft side, Teams lets administrators allow only specific external domains, or block chosen ones, under External access in the admin centre. Microsoft says that once you set an allow list, all other domains are blocked. It won't stop a scam that comes through WhatsApp, but it reduces the number of unknown parties who can contact staff directly in Teams.

For high-risk requests such as help-desk password resets or access changes, Microsoft Entra Verified ID offers Face Check. Microsoft describes it as matching a user's real-time selfie against the photo on a verified credential, with a liveness check, and returning a confidence score. Microsoft says Face Check is included in the Microsoft Entra Suite (S$185.00, listed 25 Oct 2026). It is a tool for identity verification workflows, not a filter that sits on ordinary meetings. Weigh it against simpler process controls first.

What should an employee do during a suspicious call?

Give staff a short script they can use without feeling they're being rude to a senior person:

  1. Pause. Say you will need to follow the payment process and will call back. A genuine executive will accept this.
  2. Don't act on the call. Don't transfer funds, share personal details or open attachments sent during or after it.
  3. Verify separately. Contact the person through the company directory or an existing, known channel.
  4. Escalate. Tell your manager and IT security, and keep the messages and meeting link.
  5. Report fast if money moved. The joint advisory says to call your bank immediately to block transactions and make a police report. The ScamShield Helpline is 1799.

Leaders matter most here. Staff will only challenge a "CEO" on a call if leadership has said, in writing and in person, that challenging is expected and won't be held against them.

Does meeting-room technology help or hurt?

Collaboration tools aren't the weak point in these cases. The weak point is a payment process that trusts what someone sees and hears. Still, managed tools help. A company-controlled Teams or Zoom tenant with sensible external access settings is easier to police than staff joining meetings from personal accounts. Securing the room devices themselves is covered in Another Cyberattack in the Headlines. But Is Your Meeting Room Secure?

There is also a link to recording and AI policies. Deepfakes rely on source audio and video, so it's worth controlling who can record internal meetings and where those recordings end up. See Does Your Company Need a Policy Against AI Recording Devices? and Someone in Your Meeting Could Be Wearing AI Smart Glasses. Would You Know?

What to do next

  • Write a one-page rule: no payment or bank-detail change is authorised by a call, video meeting or chat message alone.
  • Set dual approval thresholds and a call-back requirement using directory numbers.
  • Agree a challenge question or code phrase process for executives and their direct reports.
  • Review Teams external access and decide whether an allow list suits your business.
  • Brief finance, executive assistants and anyone who handles payments using the SPF/MAS/CSA advisory and CSA's warning signs.
  • Run a tabletop exercise: a "CFO" calls on a Friday evening asking for an urgent transfer. Does the process hold?

Frequently asked questions

Have deepfake voices really been used in corporate scams? Yes. CNN reported that an Arup employee in Hong Kong transferred HK$200 million after a video call with deepfake recreations of senior colleagues, and The Guardian reported a failed attempt against WPP that used an AI voice clone in a Teams meeting.

Has Singapore issued guidance on deepfake executive scams? Yes. The Singapore Police Force, MAS and CSA issued a joint advisory on 12 March 2025 describing scams where executives are impersonated on Zoom calls, and urged businesses to set up protocols to verify calls and messages.

Can employees detect a deepfake by listening carefully? Not reliably. CSA lists warning signs such as poor lip sync and flat tone, but also notes that consumer detection tools are immature. Verification through a separate, known channel is more dependable.

What is the single most effective control? Make sure no call, video meeting or chat message can authorise a payment on its own. Require a call-back on a directory number and a second approver above a set threshold.

What should I do if money has already been transferred? The joint advisory says to call your bank immediately to report and block the transactions, then make a police report. You can also call the ScamShield Helpline on 1799.

SourceIT Approved

SourceIT Approved

Researched and verified by the SourceIT team for Singapore buyers.

Share This Article
Facebook

More from our blog


← Older Post Newer Post →