Cheapfakes, Robocalls, and the Future of Voice Fraud

Robocalls and cheapfakes are targeting your contact call centers, and the risk is only increasing.

Robocalls are no longer just an annoyance call about fake IRS debts or expired warranties. They’ve evolved into a serious enterprise threat, leveraging synthetic speech technology capable of deceiving automated call answering systems, can overwhelm contact centers, and trick even seasoned agents.

While headlines focus on “deepfakes” and sophisticated voice cloning, the reality is that, today, cheapfakes, utilizing low-cost, commodity synthetic voices, are already here and being used at scale. Fraudsters don’t need perfection. They just need automation, anonymity, and volume.

And that’s exactly what today’s text-to-speech engines provide.

Why Cheapfakes Are a Major Threat

As Matt Smallman points out in his recent blog post, synthetic voices don’t need to sound flawless to be effective. At a fraction of a cent per call, attackers can now generate thousands of calls that are capable of:

• Harvesting information by probing IVRs with partial data.
• Surveiling accounts by monitoring balances and deposits without leaving digital footprints.
• Exploiting systems by orchestrating multichannel fraud using synthetic voices to mislead customers and agents alike.
• Overwhelming operations by flooding automated systems with traffic that is difficult to distinguish from legitimate use.

The more enterprises rely on self-service systems, the bigger the attack surface becomes. Automation scales customer service, but it also scales fraud.

The Robocall Problem Reimagined

Traditionally, robocalls were a consumer nuisance. But as automation has grown in enterprises, so has the attack surface. In 2024, U.S. consumers were bombarded with roughly 52.8 billion robocalls, which is more than 140 million robocalls every single day.

And it’s not just irritation. The FTC reported $12.5 billion in losses to fraud in 2024, with nearly $3 billion of that tied to imposter scams, the exact kind of attack that synthetic voices enable. Losses among older adults in particular are skyrocketing, with impersonation scams costing this group $445 million in 2024, up from just $55 million four years earlier.

This shows the real shift: from noisy nuisance calls to targeted, high-cost fraud campaigns. Cheapfakes accelerate that trend by making calls faster, cheaper, and harder to detect.

The Cost to Enterprises

For organisations, the problem isn’t just fraud losses, it’s the operational cost of processing unwanted calls. Industry benchmarks put the average cost of an inbound call at around $7–$10, with higher-complexity calls in sectors like finance or healthcare reaching $25 or more.

That means:

• 10,000 unwanted calls per year could cost an organization $70,000–$250,000 in wasted agent and system time.
• At larger volumes (100,000+), the costs easily climb into the millions.

It’s clear there’s a strong ROI case for detection and prevention: stopping cheapfake-driven robocalls before they consume agent time or trick customers pays for itself many times over.

Where ValidSoft Fits In

Cheapfakes exploit automation and scale, making them hard to distinguish from genuine customer calls. Defending against them requires security that works in real time, across every voice-enabled channel, without adding friction.

That’s where ValidSoft comes in. Our Trusted Identity Platform combines next-generation voice biometrics, deepfake detection, and cryptography to secure voice interactions end to end:

• Voice-Verity®: Synthetic Speech Detection
  Every call is monitored for generative AI, computer-generated speech, and robocalls in real time, with zero customer enrollment or stored PII. Enterprises can detect and stop cheapfake attacks at the network edge before they consume agent time or compromise accounts.
• Voice Biometrics for Authentication
  Proven to be up to 10x more accurate than competitors, ValidSoft’s biometrics verifies that the caller is the real, enrolled human, not a synthetic clone. This eliminates weak knowledge-based authentication, a key cause of social engineering hacking, and creates a frictionless customer experience.
• See-Say®: Verified Voice Transactions
  By binding “who spoke” (biometrics), “what was said” (speech recognition), and a cryptographic signature, See-Say® creates an immutable record of transactions, closing gaps that fraudsters exploit in voice channels.
• Privacy by Design
  ValidSoft is the only security company worldwide granted four EuroPriSe™ privacy seals, ensuring enterprises don’t trade off compliance or trust for security.

For enterprises, this means cutting fraud losses, preventing synthetic robocalls from driving up operating costs, and enabling automation strategies without opening the door to attackers.

Defending Against Agentic AI

The rise of Agentic AI, where customers (and fraudsters) use digital assistants to interact with businesses, will blur the lines even further. Watermarking synthetic speech may help in the future, but organisations need actionable defenses today. And rest assured, dedicated hackers are capable of removing watermarking.

ValidSoft provides that defense, today. By securing voice channels against synthetic speech and robocall fraud, we help enterprises ensure their automation strategies deliver value without opening the door to attackers.

Deepfakes may grab headlines, but cheapfakes are equally insidious and pervasive. They are cheap, fast, and increasingly effective against both humans and machines. For businesses, the cost of ignoring them is measured in both fraud losses ($12.5 billion in 2024) and operational costs (up to millions annually in wasted call-handling time).

With ValidSoft’s layered defenses, voice biometrics, behavioral analytics, and real-time fraud detection, organisations can stay ahead of this threat and protect both their customers and their bottom line.