AI scams are getting harder to spot in Ghana : How can you tell what is real?

From deepfake videos impersonating President John Mahama to fake investment promotions, artificial intelligence is giving fraudsters new ways to win people's trust. Ghanaian authorities have warned about the growing threat but how can ordinary users tell what is real?

A video appears on your phone.

It shows a familiar face speaking confidently about an investment opportunity. The voice sounds right. The setting looks convincing. There is even a promise of quick returns.

Would you know if it was fake?

For many internet users, that question is becoming increasingly difficult to answer.

Artificial intelligence can now generate or manipulate photographs, audio, video and text with a level of realism that makes traditional warning signs less obvious.

And this is no longer simply a theoretical threat for Ghana.

In May, police arrested 11 people in the Volta Region over allegations that they used AI-generated deepfake videos to impersonate President John Dramani Mahama in online fraud.

According to police, the suspects allegedly used the videos to solicit money and sensitive personal information from unsuspecting members of the public. Nine were remanded in custody while two were granted bail.

The case provides a glimpse of how artificial intelligence can be used alongside familiar forms of online fraud.

But it also raises a bigger question: if a video, photograph or voice can be convincingly manipulated, what should people trust?

Ghana has already been warned about deepfakes

Ghana's Cyber Security Authority has previously warned about the circulation of AI-generated videos, or deepfakes, promoting fake investments and unapproved medical products.

The authority said the videos had been designed to appear as though public figures, including politicians and media personalities, were endorsing the products or schemes.

It advised the public to be cautious of such material and to verify information through trusted sources.

The warning is important because deepfakes do not necessarily require sophisticated hacking.

A fraudster can instead exploit something much simpler: trust.

A familiar face can make an investment look legitimate.

A familiar voice can make an urgent request for money seem genuine.

A professional-looking website can make a fraudulent business appear established.

AI can help create that appearance.

The technology behind the deception

A deepfake is synthetic or manipulated media created using artificial intelligence.

In simple terms, AI systems can be trained or prompted to produce content that resembles a real person.

That can include a person's:

• face;

• voice;

• manner of speaking;

• photographs;

• written style; or

• apparent endorsement of a product or service.

The technology itself is not inherently criminal.

The same underlying advances are being used for entertainment, education, accessibility, translation, film

production and other legitimate purposes.

The problem comes when people use them to deceive others.

That distinction matters.

AI does not create fraud. It can make existing fraud more convincing.

A familiar scam with a new disguise

Many online scams depend on social engineering, manipulating someone into revealing information or sending money.

The fraudster may pretend to be a bank official, a relative, a government agency, an employer or an investment company.

AI gives criminals another way to strengthen that impersonation.

Instead of a badly written message, they can generate polished text.

Instead of a generic photograph, they can use a convincing image.

Instead of simply claiming to be a public figure, they can create a video that appears to show that person speaking.

The danger is not that every suspicious message is AI-generated.

It is that people may begin to trust digital evidence too easily simply because it looks or sounds convincing.

The Mahama deepfake case

The case involving President Mahama illustrates the potential impact.

Police said their Cyber Vetting and Enforcement Team carried out operations between 1 and 4 May in parts of the

Volta Region.

They said the suspects were allegedly involved in creating and circulating AI-generated videos impersonating the president.

Police said investigators recovered laptops, mobile phones, internet routers and 120 pre-registered SIM cards during the operations.

They also alleged that the suspects used the fraudulent content to solicit money and sensitive personal information online.

The police said preliminary investigations suggested the suspects could be part of a wider network producing and distributing fraudulent digital content for financial gain.

Those are allegations, and the suspects are entitled to due process.

But the arrests demonstrate that AI-generated impersonation has moved beyond hypothetical discussions and into an actual criminal investigation in Ghana.

The wider African picture

Ghana is not alone in confronting the problem.

Interpol's 2026 African Cyberthreat Assessment says artificial intelligence is enabling 55% of reported cybercrime across Africa.

The organisation says AI is making attacks faster, more scalable and increasingly sophisticated.

That figure is continental, not a measure of AI-related crime in Ghana.

There is currently no basis for saying that 55% of cybercrime in Ghana is AI-generated.

But the African figure shows why Ghanaian authorities and internet users are paying greater attention to the technology.

Interpol also says online scams are among the most prominent forms of cybercrime across the continent, with criminals increasingly using social engineering and AI to improve their operations.

Why mobile money makes the issue particularly important

Ghana's digital economy has created enormous convenience.

People can receive salaries, pay bills, buy goods, send money to relatives and run businesses from their phones.

But that convenience also means that successful deception can have immediate financial consequences.

The Bank of Ghana's latest fraud report covers fraud recorded by banks, specialised deposit-taking institutions and payment service providers during 2025.

It recorded 472 fraud cases reported in the banking sector, down from 716 in 2024.

The figures cover financial-sector fraud broadly and should not be interpreted as evidence that those cases involved AI.

That distinction is important.

There is currently no official Ghanaian figure showing exactly how many financial fraud cases were AI-assisted.

But the combination of widespread digital payments and increasingly convincing digital impersonation means

consumers need to be more careful about what they see and hear online.

Seven warning signs of a possible AI-assisted scam

There is no foolproof visual or audio test for a deepfake.

Some AI-generated material can be difficult to distinguish from genuine content.

So rather than asking only whether something "looks fake", consider the circumstances around it.

1. Someone suddenly asks for money

A familiar voice or photograph does not prove that the person actually made the request.

If a friend or relative unexpectedly asks you to send money, verify it independently.

2. You are being rushed

"Send it now."

"Don't call me."

"I need it immediately."

Urgency is a common feature of fraud because it discourages people from checking the story.

Take a moment.

3. The request is unusual

Ask yourself whether the request makes sense.

Would this person normally ask you to send money to that particular account or mobile-money number?

Would they normally communicate in this way?

4. You are asked for confidential information

Never provide your PIN, password, one-time password or other sensitive security information simply because someone claims to represent a bank, mobile-money provider or government agency.

5. The investment sounds too good to be true

Guaranteed profits, unusually high returns and promises of quick wealth should raise immediate questions.

Do not rely on a celebrity endorsement or an apparently convincing video.

6. The person refuses independent verification

This is one of the strongest warning signs.

If someone says you must not call them or speak to anyone else, stop.

7. The information exists only on one suspicious platform

Look for confirmation from an independent and trusted source.

If a supposed government announcement, investment opportunity or business promotion appears only on one

unfamiliar social-media account, do not assume it is genuine.

The best defence: verify through another channel

Suppose you receive a WhatsApp message from your brother saying he has been involved in an accident and

urgently needs GH¢2,000.

The voice note sounds exactly like him.

What should you do?

Do not simply reply to the same WhatsApp account.

Call his usual number.

Contact another relative.

Ask a question only the genuine person is likely to answer.

The important principle is:

Verify the person independently, not through the message that is trying to convince you.

This is particularly useful because a scammer who has compromised an account may be able to control the

conversation.

What if the scam involves a public figure?

This is where deepfakes can become particularly powerful.

A politician appears to recommend an investment.

A celebrity appears to promote a product.

A respected businessperson appears to announce a new financial opportunity.

The video may be real-looking.

But the endorsement may never have happened.

Ghana's Cyber Security Authority has specifically warned about deepfake videos being used to promote fraudulent

investments and other products.

The safest approach is to look for the information on the person's verified official channels or the official website

of the organisation supposedly behind the announcement.

Do not send money simply because a famous person appears in a video.

What should you do if you have already sent money?

The first step is to act quickly.

Contact your bank or payment service provider immediately and report the transaction.

Keep evidence, including:

• screenshots;

• phone numbers;

• account or mobile-money details;

• messages;

• links;

• transaction references;

• photographs or videos; and

• voice notes.

Do not delete the conversation.

Do not send more money to someone who claims they can recover your funds for an upfront fee.

And report the incident to the relevant authorities.

The Bank of Ghana's Investigation and Consumer Reporting Office handles complaints involving financial

institutions and has responsibility for consumer protection and financial-sector fraud reporting.

Can AI help fight AI?

There is another side to this technology race.

Financial institutions, payment companies and cybersecurity teams can use automated systems to identify unusual behaviour and potentially fraudulent activity.

The Bank of Ghana's financial-sector framework includes fraud monitoring and collaboration with security agencies.

Its latest fraud reporting system provides a picture of the types and scale of fraud being recorded across regulated financial institutions.

Interpol is also calling for greater international cooperation as criminals increasingly operate across borders and use technology to scale fraud.

But technology alone cannot solve the problem.

A sophisticated fraud-detection system may flag a suspicious transaction.

It cannot always prevent a person from voluntarily sending money because they genuinely believe the person

asking for it is someone they trust.

That is why human verification remains important.

The problem is not just the technology

It is easy to focus on deepfakes and forget the person on the other side of the screen.

Someone who loses money to an AI-assisted scam may feel embarrassed or even afraid to report it.

But fraudsters are deliberately trying to exploit trust, urgency and emotion.

Being deceived does not necessarily mean someone was careless.

As AI-generated material becomes more convincing, even cautious users may find it harder to distinguish genuine content from manipulated content.

The answer is not to stop using technology.

It is to become more deliberate about what we believe online.

What Ghanaians should remember

The most important lesson is simple:

Seeing is no longer always believing.

A photograph can be manipulated.

A voice can be cloned.

A video can be fabricated.

A message can be generated automatically.

And a professional-looking website can still be fraudulent.

That does not mean everything online is fake.

It means that important decisions – particularly decisions involving money or sensitive personal information –

should not be based on appearance alone.

Before you act, pause.

Before you send money, verify.

Before you share sensitive information, check who is asking and why.

And if something feels urgent, emotional or unusually good, give yourself time to investigate.

STOP. VERIFY. DON'T SEND.

In Ghana's increasingly digital economy, that three-step rule may become one of the simplest and most valuable cybersecurity habits.