For years, enterprises have invested heavily in securing email, endpoint devices, and network infrastructure. Those defenses are improving, and attackers are adapting. Increasingly, they are shifting to a channel that remains fundamentally exposed: voice.

Unlike email or messaging, voice happens in real time. There’s no chance to check a link, involve IT, or step away to think through it. Responses are immediate, and that’s exactly what makes it effective for attackers.

Now, AI is amplifying that risk.

AI is making voice fraud more convincing and more scalable

Voice cloning and deep-fake technologies are advancing rapidly. With only a short audio sample, as little as ten seconds, it is now possible to replicate a person's voice with high accuracy. Combined with caller ID spoofing, attackers can convincingly impersonate executives, colleagues, or trusted brands.

Voice-based social engineering is increasingly appearing inside enterprises, not just in consumer scams, with attacks targeting employees, partners, and even supply chains.

Voice works because it feels immediate. It’s hard to ignore. Tone and urgency come across quickly, and a caller who sounds familiar or credible can push someone to act before they stop to question it.

These incidents aren’t driven by carelessness; they reflect how people respond under pressure.

Why existing protections fall short

Industry frameworks like STIR/SHAKEN were a step in the right direction; they help detect and prevent caller ID spoofing during transmission. But these frameworks address only one part of the problem.

Network authentication can confirm that the call path hasn’t been tampered with. It doesn’t indicate whether the caller is legitimate, authorized to represent a given business, or ultimately trustworthy. For example, a call may pass authentication checks yet still come from a bad actor using a legitimately assigned number.

Branded calling strengthens trust by helping consumers recognize who is calling. When backed by network-level validation, it provides a more reliable way to establish trust at the moment of interaction.

The result is a growing gap: communications that appear legitimate but aren’t.

When applied correctly, AI can also help restore trust

The same technologies enabling more sophisticated fraud can also help enterprises rebuild trust in voice communications. In practice, that shift is already starting to happen. But doing so requires moving beyond static validation toward real-time intelligence. A few areas are already starting to stand out:

1. Advanced vetting at scale: AI can check whether a number is being used legitimately, confirm that a business is real and allowed to place calls, and even validate network traffic as well as things like logos or brand identifiers. Instead of a one-time check, validation happens continuously. This can be best combined with other factors, such as email/rich messaging channels to establish additional trust factors.
2. Real-time risk assessment: Rather than relying on fixed rules, AI looks at how a call behaves: how often it’s dialing, when it’s happening, and whether anything seems out of the ordinary. This context helps determine whether a call should go through, be labeled, or be blocked. Patterns like repeated attempts or unusual timing are difficult to track manually, especially at scale in real-world environments. For incoming calls into enterprises, AI can be used to compare previously standard baseline behaviors for incoming calls from that supplier or another part of the business, and flag anomalies.
3. Caller authentication: The biggest shift happens during the call itself. AI can also be used to challenge the caller to confirm specific information available only to a legitimate company, as well as ensure they are human.

Trust must be established in real time

The implication is straightforward: Trust can no longer be established before or after a communication. It must be established during it and maintained throughout. To do so, enterprises need to treat communications as an integrated system that brings together identity, security, and customer experience.

It also requires recognizing that technology alone is not enough. Even fully authenticated calls may be ignored if they do not align with customer expectations around timing, relevance, and consent. In practice, trust is both a technical and behavioral challenge. What should enterprises do now?

To address this shift, enterprises should focus on three priorities:

1. Secure the channel end-to-end: Protect both outbound and inbound communications. Ensure that your organization's calls are verifiably legitimate while also protecting employees and systems from impersonation and social engineering attacks. This includes extending protections beyond caller ID to validate identity, authorization, and intent.
2. Embed AI into communication workflows: AI should not be an add-on. Integrate AI into how customer interactions are managed, from er vetting and routing to reputation management and engagement optimization. For example, AI can help determine when to make calls, how often to reach out, and which approaches are most likely to result in genuine engagement.
3. Move beyond defense to enable growth: Avoid viewing AI as simply a defensive tool. AI can also help scale customer engagement, automate support, and unlock new service models that would be difficult to deliver manually. Moreover, AI can proactively safeguard your brand identity, making it more difficult for bad actors to undermine your reputation.

The future is identity-driven communications

Over the next few years, communications will shift toward a model where identity, context, and trust signals are built directly into each interaction. This shift will extend across voice, messaging, and other digital channels, creating a more consistent, secure experience.

For telecom providers, this represents an opportunity to differentiate on trust. For enterprises, it will redefine customer engagement. And for users, it may finally restore confidence in a channel that has become increasingly uncertain.

AI did not create the trust problem in voice, but it is accelerating it.

The organizations that succeed will be those that build real-time trust capabilities faster than attackers can break them.

We've featured the best encryption software.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit