AI Deepfake Teams Meeting

AI Deepfake Teams Meeting Results in Over $25 Million Loss

February 6, 2024||

In a recent live meeting, a groundbreaking cybercrime unfolded where the victim was the sole human participant. As reported by the South China Morning Post, the incident marked the first of its kind in Hong Kong, and potentially globally. The incident involved the theft of a substantial sum, exceeding $25 million, through the use of AI-driven deepfake technology.

Now dubbed as “Everyone Looked Real”, this sophisticated attack occurred during a multi-person Teams Meeting, with a financial department employee being directly manipulated by an artificial version of their CFO to transfer funds to other Hong Kong bank accounts.

The evolution of AI has propelled cybercrime into uncharted territory, exemplified by these “Mission Impossible” type scams.

Originating from a seemingly innocuous phishing message, the plot unfolded with meticulous planning. Despite initial suspicions from the victim, the elaborate scheme solidified when they joined the scheduled video call, encountering what appeared to be genuine interactions among known individuals instructing them on fund transfers.

As of now, the possibility of recovering the funds remains uncertain, complicated by the company’s delayed realization of the fraud over a week later. The diminishing chances of retrieval underscore the urgency of prompt detection and response to cyber threats.

The Hong Kong police have yet to disclose specific details about the involved company or employee.

Similar incidents have involved thieves obtaining stolen employee ID cards or using publicly available images or movies to set up their scam. The advent of AI presents both unparalleled benefits and risks, with this attack serving as a prelude to scammers leveraging AI’s latest capabilities. The imperative to “harden” users becomes evident, emphasizing constant vigilance, verification, and training among staff in the realm of cybersecurity.

To further underscore the need for ongoing user training, a fundamental cybersecurity rule has cautioned against initiating fund transfers through email due to their susceptibility to forgery. The paradigm shifted with AI’s ability to mimic voices based on sample data, enabling the impersonation of individuals.

In 2022, the beginning of AI driven voice scams, over 5,000 victims fell prey to losing $11 million. The losses, though initially small, signaled criminals’ foray into AI voice driven criminal activities. The cyber criminal’s strategies evolved with the thieves learning ways to leverage the fast-coming latest AI advancements, with this recent crime surpassing the combined losses of the entire previous year from a single incident.

As the perpetual battle between cyber protectors and thieves intensifies, the incorporation of AI has elevated the conflict to unprecedented levels, emphasizing the necessity of ongoing education and adaptive cybersecurity practices. To protect against machines, we all must be smarter humans.

The cornerstone of cybersecurity has been, and still is, hardening your users, or ‘user hardening’. This is a term being thrown out often, but is crucial to protecting your livelihood. What exactly is meant by hardening? It means turning your biggest risk – your staff falling victim to a scam – into your biggest defense. So what can you do?

Human Firewall: A firewall is a piece of equipment that protects your network from the outside world the best it can. You need to have your team act as a human firewall, filtering out the good and bad in every digital interaction they have. It’s crucial to equip your staff with the right knowledge and safeguards to help identify and prevent daily threats.

  • User Training: There are systems, tools, and companies that can help train your team on the threats so they are aware of them to best avoid them.
  • Dark Web Monitoring: Most phishing and impersonation attacks occur from information found about you. While much of this can be found on the internet, there are troves of personal data that can be used to lure your team into a false sense of trust with a hacker. Or it can be used to blackmail. The good thing is that it’s relatively easy to find what is out there about you so you can defend against it.
  • Web Filters: Typically called a DNS filter, these tools block malicious web traffic at its source, only allowing your team to go to good, or authorized, sites. This helps cut down on the success rates of hackers making impersonation websites or login pages that your team falls for.
  • Two-Factor Authentication: EVERY account you have that can have multifactor authentication (also called two-factor authentication or 2FA) should be enabled. Think of a password like closing the door to your house, and multifactor authentication as locking your door. It should always be enabled; yes, it can be pain, but it is one of the strongest defenses you can have.
  • Password Vault: DO NOT share passwords. EVER! This is a hacker’s dream. Even if it is for something you think is not important, gaining access to your Amazon or Walmart account can tell a hacker much about you, that can be used to impersonate you to your co-workers. They saw you bought a new dress shirt – that can be used to email a co-worker asking, “Hey what did you think about that green shirt I wore yesterday?” Now they have instant credibility, and your co-worker is none the wiser!
  • Safe Word: Just like having one with your kids to ensure they are safely picked up from school, this safe word keeps your money safe. It may sound silly, but with the capabilities of AI, having a word you can use with someone when it relates to financial transactions helps insure you are not speaking to an imposter.
  • Encryption Keys: One of the latest upcoming protection measures to AI driven cybercrime is the use of an in-person exchange of encryption keys.

There are also many, many technical tools and buzzwords around cybersecurity that can be implemented. Most of these tools do require professional technologists to implement and manage. What the hackers don’t want you to know though is a vast majority (typically 70%+ depending on source) of attacks can be caused or prevented by them. By having a strong concept of Human Firewalls, then you can be well on your way to keeping your data safe.

Should you ever have any questions relating to cybersecurity, looking to have an expert assist with training your team to be human firewalls, or are looking for full-stack cybersecurity solutions, give us a call today!

More from Steve...