Hiya AI Phone App Protects Against Deepfakes, Other Scams – Source: securityboulevard.com

Voice intelligence and security firm Hiya is rolling out an AI-powered call assistant mobile app aimed at protecting users from fast-growing deepfakes, scams, robocalls, and similar mobile threats.

The Seattle-based company’s Hiya AI Phone app answers calls that are unknown to the user, asks the callers to state their names and purpose, and then analyzes and decides whether to connect the calls to the app user. It also will connect the call and state the caller’s name and purpose and let the app’s owner decide whether to call back.

The app, which is a available in both Apple’s App Story and Google’s Play store, is more than a convenience, according to Hiya executives. It also is designed to protect against threats like deepfakes, which are fueled by the rapid rise in the adoption of generative AI.

The app can detect AI-generated voices and deepfakes – which are made to sound like real people – by using AI models trained on the user’s calls in real-time to detect patterns in the audio that show the voice likely is AI- generated or modified. The app then alerts the user by vibrating the phone and displaying a warning.

Deepfakes on the Rise

Cybercriminals can more accurately run deepfake voice and video scams in such campaigns as phishing scams and identity theft. Henry Patishman, executive vice president for online identity verification (IDV) solutions at Regula, an identity security company, wrote in a report that last year half of all businesses experienced fraud that involved audio and video deepfakes and 66% of business leaders surveyed said deepfakes are serious threats to their companies.

Businesses lost an average of almost $450,000 to deepfakes, according to the company.

“Deepfakes are increasingly being used in fraudulent activities, and concern is growing as the technology becomes more advanced and affordable,” Patishman wrote. “Criminals can now easily create convincing fake identities and perform presentation attacks, such as displaying a deepfake on a screen during verification. This presents a serious challenge for online identity verification (IDV) across many industries.”

Scam Protections and Call Transcripts

Hiya AI Phone also offers protections against other scams and provides automated note taking that can be used in summaries or transcripts of a conversation. In detecting scams, the app continuously analyzes calls in real-time by using AI model trained to detect scam language and similar indicators within conversations.

Like with deepfakes, the app will alert the user if it identifies a potential scam from a person, recording, or an AI-generated voice via a vibrating phone and displaying an on-screen warning.

The app also automatically transcribes and summarizes call conversations using AI models trained by Hiya. The call data is encrypted and is accessible only via the user’s device.

“For years, phone apps have stagnated, offering little innovation since their introduction on smartphones,” Hiya founder and CEO Alex Algard said in a statement. “Mobile phones are the ideal platform for AI to make a meaningful impact, and transforming the call experience with a digital assistant that works seamlessly for you on your calls is the perfect starting point.”

The app is a free download for Android and iPhone users, according to Hiya. Premium features are available via a $9.99 per-month subscription with a free two-week trial available.

A Good Step

Eyal Benishti, founder and CEO of email security software maker Ironscales, is cautiously optimistic about Hiya’s new app.

“As tools used to perpetrate deepfake scams become more sophisticated and accessible, we’ll surely see a significant uptick in these types of attacks, and they will only become more realistic and more convincing over time,” Benishti said. “While an iPhone app tackling this issue is a promising step forward, the real challenge lies in ensuring cybersecurity can evolve as quickly as the threat itself. It’s not enough to rely on automated AI defense technology; continuous refinement and a human-in-the-loop approach to AI-enabled defenses are essential to protect users effectively.”