1 in 4 calls reviewed by Hiya contain AI-generated audio

One of the biggest emerging threats for voice calls is the rising use of AI-generated audio.

While AI-generated audio can be used for legitimate purposes — such as appointment reminders or emergency alerts — it’s increasingly being exploited by criminals to power deepfake voice scams. Scammers are now using AI to:

• Automate tasks, such as creating specific scripts based on the call recipient's location.
• Remove foreign accents in scam calls.
• Create voice clones of real people to make scams more convincing. (See examples below).

How common are AI-assisted scams?

Hiya has a powerful tool for tracking scam and nuisance call trends: the Hiya honeypot. This system consists of more than 100,000 Hiya-owned phone numbers specifically designed to analyze unwanted calls.

In Q1 2025, between January 10 and March 20, Hiya’s audio intelligence team used Hiya’s AI Voice Detection product to run thousands of honeypot calls through its audio deepfake detector. They ultimately captured 8,455 calls with sufficient speech for analysis. The results:

• 25% of calls in the honeypot contained AI-generated audio.

Using a subset of the AI-generated calls, the team conducted a deeper analysis to identify the intent behind each one, finding:

• Of the calls that were AI-generated, 55% were identified as fraud — scams that used audio deepfakes to deceive recipients — while 35% were classified as nuisance, annoying perhaps, but legal and relatively harmless. The intent of the remainder could not be determined or were mistaken calls to wrong numbers.

AI-generated calls are more likely to be fraudulent

The data reveals a troubling new trend: the rise of deepfake scams. Calls using AI-generated voices are significantly more likely to be fraudulent, with bad actors leveraging synthetic audio to deceive and exploit people. While AI tools are accessible to both good and bad actors, we're now seeing AI-generated voices in many fraud and nuisance calls.

This marks the first time Hiya has tracked the presence of AI in our U.S.-based honeypots — an important step as we monitor whether the use of deepfake scams is increasing or stabilizing over time.

Here are just a few examples of AI-generated scam calls captured in the Hiya honeypot in Q1:

• Google business listing scam
• Cable TV discount scam
• Debt management scam
• Tax relief scam

Examples of deepfake voice-clone scams

The most sinister of AI-generated calls are deepfake voice-clone scams. That’s where a fraudster takes a sampling of a real person’s voice — often pulled from social media — and manipulates the audio using AI technology to pull off a scam. Here are a few recent examples:

• How con artists are using AI voice cloning to upgrade the grandparent scam | Canadian Broadcasting Corporation
• Hong Kongers lose 870 million baht to scams in a week, AI voice-cloning used | Bangkok Post
• Cloned customer voice beats bank security checks | BBC
• I'm a security expert, and I almost fell for a North Korea-style deepfake job applicant …twice | The Register
• Wichita mother nearly duped by AI voice cloning scam | KWCH-TV

How to fight back against deepfakes and identity AI-generated audio

How can carriers protect their subscribers, businesses protect their employees, and individuals protect themselves from deepfakes? The Hiya AI Phone mobile app, introduced earlier this year, is the first AI call assistant that stops deepfake scams in real time. Key features include:

• AI Voice and Deepfake Detection – The app detects AI-generated voices — live and recorded — by analyzing subtle audio patterns and notifies users when an AI voice is detected.
• Real-Time Scam Protection – The app continuously analyzes call audio to detect scam language and suspicious patterns.
• Intelligent Call Screening – An AI assistant answers unknown calls, asking callers to state their name and reason for the call, and blocks scam and spam calls while connecting legitimate ones.
• Call Summaries & Transcription – Conversations are transcribed and summarized using AI models tuned by Hiya. Data is encrypted and remains accessible only to the user on their device.

With the rise of deepfake voice scams and BYOD policies that allow employees to use their personal mobile phones for work, it's more critical than ever for businesses to secure employee communication. The Hiya AI Phone app helps safeguard employees from voice-based fraud and social engineering attacks — while boosting productivity through smart call screening and prioritization.

Even better: the technology behind the Hiya AI Phone is available to Hiya partners to integrate into their own apps, devices, or network-based services — ensuring flexible, scalable protection across any environment.

Learn more about the Hiya AI Phone app and to download a free two-week trial.