By the SightShield team · · 3 min read
When you cannot trust a familiar voice
Cheap AI tools can now clone a voice from a short recording and generate convincing video of a face. Criminals use them to impersonate managing directors, finance heads, vendors and family members — asking for urgent payments, gate access, goods release or confidential information.
Real cases
- India, October 2024: Sunil Bharti Mittal said fraudsters cloned his voice and called a senior executive in Dubai asking for a large fund transfer. The executive recognised the fraud and no money was lost.
- Hong Kong, reported February 2024: a finance employee made 15 transfers totalling about HK$200 million after a video conference in which the "CFO" and colleagues were deepfakes.
- Kerala, 2023: police reported a man losing Rs 40,000 to a video call using a former colleague’s face; the bank froze the amount.
The scale of cyber fraud
Government data reported to Parliament put losses from cyber fraud in India at about Rs 22,845 crore in 2024, across all types of fraud. Deepfakes are a small but growing part of that, and businesses are attractive targets because a single approval can move a large sum.
Why Cheap Wi-Fi Cameras Are a Security Risk →Verification rules that work
- No payment, goods release or access change on a phone, WhatsApp or video call alone — always confirm through a second, known channel.
- Call back on a number you already have, not one given in the message.
- Use code words or internal reference numbers for urgent requests.
- Two-person approval for payments above a set amount.
- Train finance, admin, reception and gate staff — they are the usual targets.
A simple payment verification flow
| Step | What happens |
|---|---|
| 1. Request received | By call, video, WhatsApp or email — treat every channel as unverified |
| 2. Pause | No payment or change on the same call, however urgent |
| 3. Call back | On a number from your own records, not one given in the request |
| 4. Second approval | A second person approves payments above a limit or any bank-detail change |
| 5. Record | Log the request, the call-back and the approval |
Physical controls still matter
Some scams aim to get someone through the door or a vehicle out of the gate. Physical controls give you something AI cannot fake easily:
- Access control on offices, stores and server rooms — a cloned voice cannot open a door.
- Visitor management at reception with ID checks and host confirmation.
- Gate procedures that require a written gate pass for goods leaving the premises.
- CCTV at gates and reception, so you have original footage of who actually came in.
If you have been targeted
- Call the national cyber crime helpline 1930 immediately, especially if money has been transferred — speed matters for freezing funds.
- Report the incident on the National Cyber Crime Reporting Portal, cybercrime.gov.in.
- Inform your bank and ask them to block or recall the transaction.
- Preserve evidence: call logs, messages, recordings, emails and CCTV footage.
- Warn colleagues and clients if your name or a senior person’s voice was used.
Quick answers
How can I tell if a voice call is a deepfake?+
Often you cannot. Rely on process instead: verify any unusual request through a separate known channel before acting.
Are deepfake scams happening in India?+
Yes — including a widely reported 2024 attempt using Sunil Bharti Mittal’s cloned voice, and police-reported deepfake video-call fraud.
How does physical security help against deepfakes?+
Access control, visitor checks, gate passes and CCTV verify people in person, which a fake voice or video cannot get past.
What is the cyber fraud helpline number in India?+
Dial 1930, the national cyber crime helpline, and report the incident on cybercrime.gov.in.
Should we stop using voice calls for approvals?+
No, but never treat a voice or video call alone as proof. Always verify through a separate channel and a second approver.