AI is giving phishing attacks an edge – here’s how to fight back

Artificial intelligence (AI) is transforming our world – and cybercriminals are taking full advantage. Phishing, long a favourite tactic in the cybercrime playbook, has now been supercharged by AI, making it harder than ever to detect.

Even well informed and experienced individuals are being fooled by AI-enhanced scams that mimic real messages with alarming accuracy. And in South Africa, the impact is already visible.

According to recent Kaspersky research, 67% of local businesses experienced phishing incidents in the last 12-months.

This marks a sharp increase, with 29% more cyberattacks reported overall. What is more concerning is that 53% of local cybersecurity professionals expect AI-driven phishing attacks to grow significantly in the year ahead.

The arms race is no longer just between attackers and defenders – it is between machines, too.

Why phishing is harder to spot now

Historically, phishing attacks were easy to flag. For instance, these had poor grammar, generic messaging, or strange sender addresses that gave them away.

But those tells are disappearing. With the help of generative AI and machine learning, cybercriminals can now generate emails that mirror a company’s tone, branding, and even recent events – creating highly targeted scams.

AI does not just personalise; it scales. It can sift through thousands of public records, social media posts, and company updates to craft tailored messages for each recipient – from interns to CFOs.

Deepfake technology adds another layer of deception, with scammers able to impersonate executives in video calls or voicemails, urging staff to transfer funds or share credentials.

Traditional defences are not enough

Most corporate email filters were built to detect conventional threats, not AI-engineered messages. Criminals can now use AI to mimic safe email structures, tricking systems into letting dangerous content through. In fact, machine learning lets attackers test and refine these emails at scale – until they find what works.

While awareness training has helped, AI phishing campaigns are still successful as they exploit psychology – urgency, fear, trust in authority – to override critical thinking.

Even experienced users can be caught off guard when a message looks exactly like it came from the CEO.

A smarter, multi-layered defence

So how can South African businesses respond?

AI must be met with AI. Kaspersky’s AI and Cyber Defence global study found that 94% of cybersecurity professionals now see AI-enhanced tools as essential to defending against modern threats.

But AI alone is not enough. The solution lies in combining smart tools with skilled people and resilient processes.

Here are four key steps businesses should consider:

1. Modernise your security stack

Tools like Kaspersky Next and Kaspersky Security for Mail Server use AI to detect suspicious patterns, prevent phishing attempts from reaching inboxes, and adapt to emerging threats in real time.

These platforms can spot subtle anomalies – like slight changes in phrasing or metadata – that a human might miss.

2. Build a Zero Trust environment

Zero Trust assumes no user or device should be trusted by default, even within the network perimeter. By limiting access to sensitive data, businesses can contain damage if a phishing attack succeeds.

Layered authentication, micro-segmentation, and real-time monitoring help enforce this model.

3. Train for the AI age

Security awareness training remains vital – but it must evolve. Kaspersky’s Automated Security Awareness Platform is an example of a solution that continuously educates employees on identifying AI-enhanced threats.

Adaptive learning ensures that staff stay alert as attack methods evolve.

4. Close the skills gap with partnerships

With 50% of organisations globally citing a shortage of cybersecurity talent, and 39% struggling to find the right expertise, many South African businesses are at risk simply because they cannot hire fast enough.

Managed security services, like those offered by trusted cybersecurity providers, can bridge this gap and deliver enterprise-grade protection without the overhead.

The road ahead

AI is changing the threat landscape. Yet this same technology can also empower defenders. With AI-assisted threat hunting, real-time behavioural analysis, and faster incident response, cybersecurity teams can stay one step ahead.

The key is to act now – before AI-driven phishing causes extensive damage.