Realistic and convincing deepfakes are becoming fraudsters’ weapon of choice, so organisations must future-proof their risk mitigation plans against this emerging threat.
In early 2024, a Hong Kong-based employee of the UK engineering firm Arup transferred a total of US$25m to five separate bank accounts, before discovering that they had fallen victim to an elaborate fraud. The employee had been duped into making the payments as a result of a conference call in which all other participants, including someone purporting to be the company’s chief financial officer, were deepfakes — their voices and likenesses had been digitally cloned and manipulated with AI by fraudsters.
Although the scam heralded a new frontier in crime, Michelle Chia, chief underwriting officer for cyber, Americas, at AXA XL Insurance, a leading provider of insurance for businesses worldwide and part of the global insurance group AXA, stresses that this was an old crime committed with new technology. “Employees often receive emails purporting to be from their CEOs, requesting they move money into a bank account, which the vast majority have learned to ignore. This fraud was the same type of crime, but rather than email, it was conducted via a video call employing new techniques that capitalise on a lack of vigilance caused by perceived technological limitations. What is worrying is that deepfakes are now more realistic — and harder to detect — and that is only going to get worse,” says Ms Chia.
Deepfakes are not limited to impacting an organisation financially; they can also be used to destroy reputations. For example, deepfakes can be employed to imitate a company’s CEO saying bad things about their products, organisations or competitors. What is worse, reputational attacks can be carried out without breaching a company’s firewall.
“Fake news is a powerful weapon that adversaries can monetise without compromising an organisation’s multi-factor authentication controls. They just have to do something that causes its share price to fall,” explains Ms Chia.
Preparing for deep trouble
Companies need to plan for deepfake risk mitigation, according to Vanessa Leemans, AXA XL’s head of cyber, UK & Lloyd’s. “Organisations can better prepare themselves before, during and after a deepfake attack, so they are better able to identify deepfakes, understand how to respond to a deepfake attack, and how to recover. But they must create and test these plans now,” she warns.
Employees are a vital part of the identification process. Most organisations have in place training to combat phishing emails. Ms Leemans recommends that this training should also include deepfakes. “If all employees are trained to detect and report deepfakes, an organisation broadens the protective net it can cast.”


The same practices in general cybersecurity also apply to managing deepfake exposure. Zero-trust user validation processes are especially useful to prevent deepfake video calls. Multi-factor authentication to access a video call stops adversaries in their tracks.
Although software is an important tool in the fight against deepfakes, it cannot prevent all deepfake attacks, especially reputational attacks. Ms Leemans explains that a deepfake attack is like any other PR threat. “Organisations must develop a clear, concise message about what has happened, and quickly share the company's position to nip it in the bud.”
The future of fakery
While deepfakes are becoming harder to spot, their future impact remains uncertain. As with other cybersecurity risks, attackers and defenders are on a constant treadmill. When defenders improve their detection methods, adversaries improve their deepfakes. “The question is not how fast you run, but how fast you run relative to attackers,” Ms Chia advises. “Sometimes defenders will be winning. Other times, it will be the attackers. One is always chasing the other. It's an arms race.”
It is almost impossible to anticipate what the next fraud trend will be, so the best organisations can do is to proactively manage risk. However, even the best-designed programmes cannot mitigate risk entirely. This is where insurers like AXA XL step in.
Ms Chia believes that the perpetrators of many types of fraud will start to use deepfakes. She cites the example of the recent, eye-catching incident that targeted a luxury food retailer, where fraudsters used fake credentials to steal £300,000 worth of product. This is just the type of fraud that deepfakes could make more convincing. “We offer crime policies to cover this type of loss. The question is whether a firm has taken appropriate precautions to make sure they have identified where they might suffer financial loss, and how they manage that. The insurance industry works hard to keep up with the evolving risks our clients face, and provide the risk transfer our clients need.”
But, as her colleague Ms Leemans is keen to point out, insurers do not just provide a risk transfer service. AXA XL monitors what happens across a broad portfolio and identifies new trends as they occur. As a consequence, they can recommend to their partners on the most appropriate risk mitigation factors and controls to help them to reduce their risk exposure. “In the face of rising risk from deepfakes, cyber insurers play an important role by sharing insights to help organisations improve their risk awareness and protect themselves from these evolving risks,” she explains.
The emergence of deepfake fraud could herald a period where adversaries run faster than defenders, who must play catch-up. That race cannot be won alone — it will be a team effort.
Explore more

Facing the future: How embracing the emerging risks can drive innovation and growth


