Μετάβαση στο περιεχόμενο

Το καλάθι σας

Το καλάθι σας είναι άδειο.

Προσθέστε μια υπηρεσία από το κατάστημα και θα σας περιμένει εδώ μέχρι την αγορά.

Δείτε το κατάστημα

Μετά την πληρωμή

  1. Η ομάδα διαβάζει τα στοιχεία σας και σας στέλνει email, συνήθως μέσα σε μία εργάσιμη ημέρα.
  2. Παρακολουθείτε τη δουλειά και μιλάτε με την ομάδα στην πύλη πελατών.
Μερικό σύνολο

Εφάπαξ χρεώσεις υπηρεσιών, σε USD.

Ταμείο Δείτε το καλάθι
Απειλές ΤΝ

Deepfake CEO calls: the video meeting that cost HK$200 million

Criminals now clone the voices and faces of executives to order payments. The real cases, and the callback and approval rules that stop them.

Δημοσίευση: 5 λεπτά ανάγνωσης7 πηγέςΤα άρθρα είναι στα αγγλικά.

In January 2024, a finance employee at the engineering firm Arup in Hong Kong joined a video call with the company’s chief financial officer and several colleagues. They asked for a confidential set of payments. The employee made 15 transfers to five bank accounts, totalling HK$200 million, about £20 million.

Nobody else on that call was real. Arup later confirmed: “We can confirm that fake voices and images were used.” The faces and voices were generated by AI.

It is not one case

WPP. Criminals set up a WhatsApp account with a public photo of WPP’s chief executive, Mark Read, and used it to arrange a Microsoft Teams meeting with a senior executive. In the meeting they played a voice clone of Read with YouTube footage, and typed as him in the chat. They were after money and confidential information. Staff spotted it. Read’s warning to his leadership: “Just because the account has my photo doesn’t mean it’s me.”

Ferrari. In July 2024, an executive received WhatsApp messages that appeared to come from chief executive Benedetto Vigna, from an unknown number. A call followed, in a voice that mimicked Vigna’s southern Italian accent, pushing an urgent, confidential deal. The executive asked one question: the title of a book Vigna had recommended a few days earlier. The caller could not answer and hung up.

Arup’s own warning. After the Hong Kong loss, Arup’s global chief information officer, Rob Greig, said: “What we have seen is that the number and sophistication of these attacks has been rising sharply in recent months.”

What the FBI says

The FBI has warned about this directly.

  • In December 2024, it said criminals use AI to “generate videos for real time video chats with alleged company executives”, and that generative AI “reduces the time and effort criminals must expend to deceive their targets.”
  • In May 2025, it warned that AI voice clones “can sound nearly identical” to the real person.
  • For the first time, the FBI’s report on 2025 internet crime counts losses tied to AI: about $893 million. That includes business email compromise using AI-written emails and voice cloning, executive impersonation, and cloned voices used to approve transfers by phone. The FBI says the figures could be understated.

Why it works

A deepfake does not need to be perfect. It needs to be good enough for a few minutes, on a call where someone senior is asking for something urgent and secret.

The criminals only need public material. Executives speak at events, appear in videos and post on social media. A few clips are enough to build a voice.

And they attack the process, not the software. The real weakness at Arup was that a video call was enough to move millions.

The rules that stop it

You cannot train people to spot every fake. The tools improve too quickly. Build rules that work even when the fake is perfect.

1. Call back, always. Any request to pay, change bank details or share sensitive data is confirmed by calling the person back on a number you already have, never on a number, link or meeting invite from the request itself. The FBI gives the same advice.

2. Agree a code word. Senior staff and anyone who moves money agree a phrase or question that never appears in email or chat. The FBI recommends a secret word to check identity. Ferrari’s executive did the same with a book title.

3. Two people approve every payment over a set amount. No single person, however senior, can order and release a large payment alone. The approver checks the request through a separate channel.

4. Secret and urgent means stop. Make it a written rule: a request that is both confidential and urgent is checked before anything happens. This is exactly the pressure the Arup, WPP and Ferrari callers used.

5. Bank detail changes wait. A new or changed supplier account is confirmed by phone on a known number and held for a set time before the first payment goes out.

6. Nobody gets in trouble for checking. Tell everyone, in writing, that the chief executive will never mind a callback. The staff who stopped the WPP and Ferrari attempts are the reason those stories ended well.

7. Lock the systems behind the rules. A rule only works if your systems enforce it. If one account can change a supplier’s bank details and approve the payment, the rule is a suggestion. Use separate permissions, MFA on every finance and admin account, and alerts on changes to payment details.

What to do this week

  • Write the callback and code-word rule on one page and send it to everyone who touches money.
  • Check who in your company can change bank details or approve payments, and in which systems.
  • Remove any path where one person can do both.
  • Turn on MFA for every finance, email and admin account.

Where EWC fits

Voice and video cloning keeps improving. Our security team follows these AI impersonation methods and builds them into how we test.

Our Tailored cybersecurity assessment tests the systems behind your payment rules: sign-in, account recovery, permissions and business logic across the websites, applications and APIs you agree with us. We find where one stolen or tricked account could change payment details or approve a transfer, and give you a prioritised plan to close it. It covers your systems, not phone calls to your staff.

Πηγές

Σε τι βασίζεται αυτό το άρθρο.

7 πηγές
  1. DezeenArup victim of multimillion-dollar deepfake video scam in Hong Kongdezeen.com
  2. CNNBritish engineering giant Arup revealed as $25 million deepfake scam victimedition.cnn.com
  3. MIT Sloan Management ReviewHow Ferrari Hit the Brakes on a Deepfake CEOsloanreview.mit.edu
  4. Marketing BeatWPP CEO's identity usurped by deepfake scammersmarketing-beat.co.uk
  5. FBI Internet Crime Complaint Center (IC3)Criminals Use Generative Artificial Intelligence to Facilitate Financial Fraudic3.gov
  6. FBI Internet Crime Complaint Center (IC3)Senior US Officials Impersonated in Malicious Messaging Campaignic3.gov
  7. iTnewsAI-assisted fraud makes big debut in FBI's cybercrime statsitnews.com.au

Πού βοηθάμε · Επίπεδο 3

Προσαρμοσμένη αξιολόγηση κυβερνοασφάλειας με πιστοποιητικό

Αξιολογήστε τους κινδύνους ασφάλειας που μετρούν για την επιχείρησή σας, με αναφορά και πιστοποιητικό.

Τι περιλαμβάνει

  • Αξιολόγηση στα μέτρα σαςΣυμφωνημένα συστήματα, κίνδυνοι και στόχοι ελέγχου.
  • Ευρήματα και παρουσίαση στην ομάδαΤεκμήρια, διαδρομές επίθεσης και πλάνο αποκατάστασης.
  • Πιστοποιητικό αξιολόγησης EWCΑντικείμενο, ημερομηνίες και στοιχεία της αναφοράς.