Fraud attempts with deepfakes have increased by 2,137% over the last three years, according to fresh research from digital identity platform Signicat.

The Battle Against AI-Driven Identity Fraud report, based on a survey of over 1,200 fraud decision-makers across the EU and UK, found that more than 42% of fraud attempts detected in the financial sector are now due to AI, with deepfakes the most common type of digital identity fraud faced by companies, accounting for one in every 15 cases.

The evolution of deepfake technology has enabled two main types of attacks to take hold.

Presentation attacks, which can include fraudsters wearing masks and makeup to spoof another person, but also where a camera films another screen showing a deepfake in real-time, target activities such as account takeovers or fraudulent loan applications.

Meanwhile, in injection attacks, malware or fake data is ‘injected’ directly into the system without passing through the device’s camera, compromising the integrity of information across communication channels.  

The steep rise in deepfake fraud is part of a broader trend of AI-driven identity fraud, as cybercriminals increasingly leverage advanced technologies to exploit financial systems, with overall fraud attempts climbing by 80% in three years, resulting in losses exceeding $362 billion (£286bn) between 2023 to 2028.

Although the majority of businesses now recognise the scale of the threat, with three-quarters looking to implement AI-driven identity fraud solutions in the next twelve months, Signicat’s report argues that most organisations lack understanding of the problem.

Illuminating this best is the 75% of respondents who felt that deepfakes would never be convincing enough to completely fool the security measures of financial institutions, despite 71% of fraud attempts using AI being successful.

Compounding this problem, fraud decision-makers said their efforts to detect and combat enhanced identity fraud are being hampered by multiple barriers, including a lack of expertise (76%), inadequate budgets (76%), and insufficient time (74%), which has led to just 22% of financial institutions having implemented AI-based fraud prevention tools.

Recommended reading

• Is It Now Practically Impossible To Identify Deepfakes?
• UK Gov Cracks Down on Explicit Deepfake Creators
• Creating Deepfakes To be Criminalised in UK

To stay ahead of the curve, Signicat’s report warns that financial institutions and businesses need to proactively strengthen their cybersecurity measures, update fraud detection systems, and improve employee and customer awareness, along with simply investing in AI solutions.

“Fraudsters are using AI-based techniques that traditional systems can no longer fully detect,” said Pinar Alpay, chief product and marketing officer at Signicat.

“A multiple detection setup is crucial. By combining early risk assessment, robust identity verification and authentication methods based on facial biometrics, and ongoing monitoring, companies can better protect both their operations and their customers.

“Orchestration of these tools in the optimal combination is the essence of a multi-layered protection”.