Fraud attacks hit record high thanks to AI and biometric restrictions
Cyber fraud has reached a new peak this year – as global economic uncertainty takes hold, artificial intelligence (AI) improves, and regulatory frameworks weaken the defenses of businesses and consumers alike.
This is the conclusion published in the Voice Intelligence & Security Report (opens in new tab) from voice tech experts Pindrop (opens in new tab). Analyzing five billion calls and three million fraud cases within the largest financial institutions, the company found that there’s been a 53% year-on-year increase in fraudulent activity in Q4 2022.
Despite the grim outlook, retail has it even worse according to the report, stating the industry is now “one of the most fraud-dense verticals”. Apparently, one in every 347 calls is fraudulent within call centers that support online retailers.
Retail under attack
But it’s not just the call centers that the fraudsters are attacking. Ecommerce and credit card fraud is also quite prevalent. In fact, recent data found that 3.6% of all ecommerce revenue in 2022 was stolen by fraudsters, while payment fraud grew 40% year-on-year between 2021 and 2022.
To help them commit this fraud, thieves are leveraging AI, mostly to generate believable deepfakes.
“In addition to new technologies, fraudsters are reverting to pre-pandemic social engineering tricks, causing retailers and financial institutions billions in losses,” said Vijay Balasubramaniyan, CEO and cofounder of Pindrop.
Regulators aren’t helping either, despite their best intentions. In some U.S. states there are laws that put extra restrictions on the use of biometrics, designed to protect user privacy (opens in new tab).
However, the residents of these states are twice as likely to experience fraud, the researchers claim. They added that “these states also contribute to 33% of all fraud losses reported in the U.S.”
The human voice is still “a vital factor” for fraud detection, but some states are allegedly working on a bill to regulate voice “without notable security of fraud-prevention exemptions,” which might render current detection mechanisms obsolete.