Pindrop receives $100 million in venture capital to expand protection against deepfakes using voice

Voice deepfake attacks, which can be used to deceive call center agents, automated systems, employees and relatives, have increased sharply this year, while the availability and quality of voice spoofing tools has increased dramatically, said Pindrop founder and CEO Vijay Balasubramaniyan. Biometric update in an interview. This is the background behind the $100 million debt financing that Pindrop received from Hercules Capital to expand its customer base for deepfake detection and voice biometrics.

The funds will also be used to develop liveness and speech recognition tools for an increasingly broad range of industries as deepfakes are used for fraud purposes in more and more scenarios.

“It used to be a cool party trick and now it’s everywhere,” says Balasubramaniyan, noting that deepfakes followed generative AI into the spotlight last year, but this year real-world attacks have picked up steam.

By combining large language models and text-to-speech engines, attackers can scale their synthetic voice fraud efforts not only based on the number of people, but also based on the computers available to them.

“For the entire last year, we had an attack on our customers about every month,” he says. “On average for the first four months, it’s now one attack per customer per day. And there are certain customers who are experiencing more attacks every day.”

That represents a 450 percent increase in the first four months of 2024 compared to the entire previous year. The rate of increase suggests that this number will be much higher by the end of the year. According to the fundraising announcement, by the end of the year, one in every 730 calls to a call center will be a scam attack of some kind.

At the same time, the technology is being used for other types of fraud, including election fraud, such as the robocall that spoofed President Joe Biden’s vote and was uncovered by Pindrop in February.

“I’ve never heard of a technology that destroys commerce, media and communications in one fell swoop,” says Balasubramaniyan. “So that’s the moment we’re in.”

The company had already raised more than $200 million through equity funding, but changed its fundraising method due to the maturity of its business, Balasubramaniyan explains.

Equity financing uses a model where one in ten successes can finance the others, whereas in the world of venture capital you can afford very few failures.

The interest rate, he says, is low compared to the massive increase in the company’s value. “It is one tenth cheaper than the increase in equity and the lack of dilution.”

Pindrop already has the scale, cash flow breakeven and positive metrics to avoid dilution, and Balasubramaniyan says the company is clear about its customer acquisition costs.

“Many publicly traded companies operate on debt because the numbers are so solid at this point,” he says. “We have matured as a company and this seems to be the most cost-effective way to raise capital.”

Both scaling Pindrop’s customer base and further product development are planned, and Balasubramaniyan notes that a very high rate of his customers are adding deepfake detection. “That rate of attachment is incredible,” he says.

Existing voice biometrics customers can also start using the technology almost immediately.

The revenue level that Pindrop achieved with its deepfake product in three months was equivalent to three years for the company’s original voice biometrics product.

John Chambers, a Pindrop investor and board member who is also founder and CEO of JC2 Ventures and former chairman and CEO of Cisco, says Pindrop developed its deepfake product exceptionally quickly and gained customers immediately, “and under Vijay’s leadership, Pindrop will achieve the fastest scale to $5 million in a completely new product space compared to any other company I know. This new investment from Hercules Capital comes at a pivotal time for Pindrop and will enable them to further develop their cutting-edge audio, voice and AI technologies and enhance their offerings to customers in the banking, finance, contact center and insurance sectors,” he says in the announcement.

Pindrop wants to take its deepfake speech recognition outside the contact center “to places like media, government and communications,” says Balasubramaniyan, “so this is a big TAM expansion opportunity for us.” He estimates the company will be more than four times as big.

The Biden robocall allowed Pindrop to show how sophisticated attackers have become. Pindrop was monitoring a single online tool for the purpose of voice spoofing when it started working on deepfakes. By the end of last year, there were already 120, and in March there were almost three times as many.

Awareness of the threat has increased with the Biden robocall, but the week-long news cycle that exposed that hoax is too long to avoid chaos if it happens again just before the November election, Balasubramaniyan warns. Public understanding continues to advance, however. “A lot of people weren’t very clear about the term liveness. Now members of Congress are talking about liveness,” he notes.

Politics must always lag behind technology, but Balasubramaniyan hopes it doesn’t get too specific about the technological requirements it makes. Watermarks, for example, can play a role in reassuring people that content is genuine. But that role is limited. In the case of the Biden robocall, only two percent of the original watermark could be detected, “and that wasn’t even an active adversary, it just went through the transmission channels.”

As more and more content is created using artificial intelligence, determining whether the person concerned has given their consent becomes a key challenge.

Pindrop’s next product could potentially be “consent detection.” In the meantime, the company has the resources it needs to serve the large and growing market demand for protection against voice deepfakes.

Article topics

biometric authentication | biometrics | deepfake detection | funding | investment | pindrop | research and development | synthetic voice | voice biometrics