Can AI Copy Your Voice, and How Can Namibian Businesses Protect Themselves?

Sep 30

10 min read

A finance employee receives a WhatsApp voice note from the managing director.

The voice sounds right. The tone sounds familiar. The instruction is urgent: transfer money to a supplier before the end of the day.

Until recently, hearing a familiar voice may have felt like strong proof of identity. That is no longer enough.

Artificial intelligence can now generate convincing imitations of a person’s voice from relatively short recordings. For Namibian businesses, this creates a new fraud risk: criminals may impersonate executives, suppliers, employees or public figures to request payments, confidential information or access to business systems.

The safest response is simple:

Do not rely on a familiar voice as proof of identity. Verify important instructions through a separate, trusted channel before acting.

This matters because AI does not create an entirely new type of fraud. It makes familiar social-engineering scams more convincing.

What Is AI Voice Cloning?

AI voice cloning uses artificial intelligence to create synthetic speech that sounds like a real person.

A system analyses characteristics such as tone, rhythm, pronunciation and speaking style, then generates new words in a voice resembling the original speaker.

Namibia’s Communications Regulatory Authority has previously warned that AI can be used for impersonation and voice phishing. CRAN notes that a small sample of someone’s voice can potentially be used to generate speech that appears to come from a trusted individual.

Recent Namibian discussion has also focused on the implications of AI voice cloning for performers, businesses and consent. A September 2026 article in The Brief highlighted that voice-cloning technology can imitate a person from a short recording and raised questions around unauthorised use and digital identity.

A business owner may naturally ask:

“Can AI really copy my voice?”

Yes. The technology exists.

The more important business question is:

“What happens if someone uses that voice to convince my employees to act?”

Can Scammers Fake a WhatsApp Voice Note?

Yes.

A voice note should no longer be treated as automatic proof that the person speaking actually recorded it.

AI-generated audio can potentially be sent through messaging platforms in the same way as an ordinary recording. A fraudster may therefore combine a cloned voice with information gathered from social media, company websites or previous communications.

Imagine an employee receives:

“Hi, I am in a meeting. Please process N$85,000 to this account immediately. I will explain later.”

The voice sounds like the owner.

The employee recognises the speaking style.

The instruction is plausible.

That combination can be extremely persuasive.

But the correct question is not:

“Does this sound like my boss?”

It is:

“Can I verify that my boss actually sent this?”

How Could Fraudsters Get Enough Audio to Copy Someone’s Voice?

Many people already have recordings of their voices online.

These may include:

interviews;

social media videos;

webinars;

podcasts;

voice notes;

company presentations;

radio appearances;

TikTok or LinkedIn videos.

This does not mean executives or professionals should stop communicating publicly.

The issue is awareness.

A senior employee with a strong online presence should assume that samples of their voice may already be publicly available. The business should therefore design payment and approval procedures that do not depend solely on recognising a voice.

A practical rule is:

Identity should be verified by process, not familiarity.

Why Is AI Voice Fraud Particularly Relevant to Businesses?

Businesses already deal with impersonation fraud.

Scammers pretend to be:

CEOs;

suppliers;

banks;

customers;

government officials;

IT providers;

colleagues.

AI potentially makes those impersonations more credible.

Namibian regulators and financial institutions have already warned about continuing vishing, or voice-phishing, scams. Vishing typically involves criminals using calls to manipulate people into transferring money, revealing sensitive information or providing account access.

AI voice cloning adds another layer.

Instead of merely claiming to be the managing director, a scammer may be able to sound like the managing director.

This is why businesses need to update an old assumption:

Recognising someone is not the same as verifying someone.

What Should I Do If I Receive an Urgent Voice Note Asking for Money?

Do not act on the voice alone.

Pause.

Contact the person independently using a number or communication channel you already know.

For example, if a WhatsApp voice note apparently comes from the owner asking for an unusual payment, call the owner directly through the normal number or follow the company’s established approval procedure.

Do not ask the suspicious sender:

“Is this really you?”

A fraudster will simply say yes.

Instead, use an independent channel.

The safest principle is:

Verify outside the conversation that created the request.

This applies equally to email, WhatsApp, phone calls and video calls.

What Should Employees Check Before Acting on a Voice Instruction?

Employees should focus less on whether the voice sounds convincing and more on whether the request itself makes sense.

Ask:

Is this payment unusual?

Has the bank account changed?

Is the person asking me to bypass normal procedures?

Am I being told not to involve anyone else?

Is there unusual urgency?

Can I verify this request independently?

A genuine executive should understand why an employee wants to verify an unusual financial instruction.

In fact, organisations should actively encourage employees to pause.

An employee should never fear being criticised for verifying a large or unusual payment.

Can Someone Clone the Voice of a CEO or Business Owner?

Potentially, yes.

Executives and public-facing professionals may have more audio publicly available than they realise.

But the solution is not to hide every recording.

The more effective response is to design internal controls around the possibility of impersonation.

A business should not have a process where:

“the boss told me verbally”

is enough to authorise a significant payment.

Important financial transactions should have independent approval or verification appropriate to the size and risk of the business.

This is particularly important in SMEs where employees may be accustomed to informal instructions from the owner.

Informality can be efficient.

It can also be exploited.

Should Businesses Create a Secret Word or Verification Question?

Some families and businesses use agreed phrases or challenge questions to verify identity.

This can be helpful as an additional safeguard, particularly for unusual or urgent instructions.

However, it should not become the only control.

Secret words can be shared, overheard or eventually compromised.

A stronger approach combines several safeguards:

independent call-back;

documented approval;

dual authorisation where appropriate;

known supplier details;

internal payment procedures;

staff awareness.

The goal is not to create complicated bureaucracy.

It is to make one convincing message insufficient to move money.

How Can Businesses Prevent AI Voice Fraud?

Create a verification rule for unusual payments

Employees should know that unusual, urgent or high-value payments must be independently confirmed.

This should apply even when the instruction appears to come from senior management.

Never bypass procedures because of urgency

Urgency is one of the oldest tools in fraud.

AI simply makes the person delivering the pressure more convincing.

A message saying:

“Do this now and do not call me”

should increase scrutiny, not reduce it.

Verify changed supplier bank details

A voice call should not be enough to change a supplier’s banking information.

Use an established contact already held by the business and verify the change independently.

Train employees using realistic scenarios

Cyber-awareness training should no longer focus only on suspicious email links.

Staff should understand that voices, images and video can also be manipulated.

A useful training scenario is:

“What would you do if you received a voice note from the managing director requesting an urgent confidential transfer?”

The discussion that follows is often more useful than a technical lecture about artificial intelligence.

How Do I Know Whether a Voice Recording Is Fake?

Sometimes there may be clues.

The speech may sound unusually flat, pronunciation may be inconsistent, pauses may seem unnatural or background sounds may not match the situation.

But businesses should not rely on detecting technical imperfections.

AI-generated media continues to improve.

A convincing fake may contain no obvious warning signs to an ordinary listener.

The better question is therefore not:

“Can I detect the deepfake?”

It is:

“Can I independently verify the instruction?”

That approach remains useful even as technology improves.

What Should a Business Do If It Suspects AI Voice Impersonation?

Do not immediately engage further with the suspected fraudster.

Preserve the voice note, number, message history, transaction request and related communication.

If money has already moved, contact the bank immediately.

Inform the person whose identity may have been impersonated.

Secure any account that may have been compromised.

Then document what happened.

If the incident is serious, involves financial loss or suggests broader compromise, appropriate technical, forensic, legal or law-enforcement support may be necessary.

The evidence may matter later.

Do not delete the recording simply because you know it is fake.

Is AI Voice Cloning Illegal in Namibia?

The legal position is more complicated than simply saying all voice cloning is illegal.

Namibia's existing copyright framework protects certain recordings and performers' rights, but recent legal commentary has noted that the law does not expressly resolve every question involving newly generated AI speech that imitates a person's voice. Contracts, consent, the source material used and the purpose of the imitation can all matter.

For businesses, the practical lesson is not to wait for every legal question to be settled.

Organisations should obtain appropriate permission before commercially replicating someone's voice and should establish internal policies governing synthetic media.

Employees should also understand that the ability to generate something does not automatically mean the organisation has permission to use it.

Why Is This a Board and Governance Issue?

AI impersonation is not only an IT risk.

It can affect:

payments;

procurement;

executive authority;

customer trust;

reputation;

confidential information;

fraud risk.

Boards therefore do not need to understand how voice-cloning algorithms work.

They do need to ask:

Could someone impersonate one of our executives and successfully authorise a payment?

Are employees allowed to challenge unusual instructions from senior management?

Do our approval controls still work if audio and video can be faked?

Who is responsible for AI-related fraud risk?

This links directly to broader AI governance.

As AI becomes more common in Namibian workplaces, governance needs to address both how employees legitimately use AI and how criminals may use the same technology against the organisation.

Namibia's current AI discussion is expanding rapidly. The country launched a National Artificial Intelligence Institute in September 2026, reflecting growing national attention to AI adoption, capability and responsible use.

Common Mistakes Businesses Make

Mistake 1: Trusting a familiar voice automatically

A familiar voice should provide context, not proof.

Mistake 2: Allowing senior people to bypass controls

Executives sometimes create fraud risk unintentionally by expecting employees to process urgent informal requests.

Good controls should apply to everyone.

Mistake 3: Training only for phishing emails

Fraud awareness now needs to cover voice, video, WhatsApp, social media and AI-generated impersonation.

Mistake 4: Trying to become expert deepfake detectors

Employees do not need to determine technically whether every recording is synthetic.

They need a reliable verification process.

Mistake 5: Blaming the employee who nearly fell for it

Modern impersonation fraud can be highly convincing.

A near miss should be treated as useful information about where controls need improvement.

This article provides general awareness and practical guidance. It is not legal, cybersecurity, technical, forensic or regulatory advice. Organisations dealing with suspected AI impersonation, financial loss, account compromise or unauthorised use of a person's voice should obtain appropriate professional advice and preserve relevant evidence.

AI can imitate a person's voice convincingly enough that businesses can no longer treat familiar audio as proof of identity.

That sounds unsettling, but the solution is practical.

Businesses do not need to identify every deepfake.

They need processes that survive one.

Verify important requests independently. Protect payment procedures. Give employees permission to question unusual instructions. Preserve evidence when something goes wrong.

Technology may change how convincing fraud becomes.

The principle remains familiar:

A familiar voice can create trust. Independent verification should create the decision.

FAQ

Can AI really copy someone's voice?

Yes. AI voice-cloning technology can generate speech resembling a real person from recorded audio. The quality varies, but modern systems can create convincing imitations. Businesses should therefore avoid using voice recognition alone to approve payments, disclose sensitive information or change important account details.

Can a WhatsApp voice note be faked with AI?

Yes. Synthetic audio can potentially be shared as a voice recording through messaging platforms. Employees should therefore verify unusual financial or confidential instructions through another trusted channel rather than assuming a familiar-sounding voice proves who sent the message.

What should I do if my boss sends a voice note asking for an urgent payment?

Follow the organisation's normal payment process and verify the request independently, particularly if the amount, bank details or urgency are unusual. Call the person using an established contact method rather than relying solely on the message that requested the payment.

How can a small business protect itself from AI voice scams?

Start with practical controls: independently verify unusual payments, confirm supplier bank-detail changes, use appropriate approval processes and train employees to recognise impersonation attempts. The objective is to ensure that one convincing call or voice note cannot by itself authorise a significant transaction.

What should I do if someone is impersonating my voice?

Preserve examples of the impersonation and related messages, inform relevant colleagues or organisations and secure any accounts that may also have been compromised. If money, reputation, confidential information or commercial use is involved, seek appropriate professional or legal advice.

Fraud Prevention. Forensic Accounting. Ethical Leadership. Digital Responsibility.