TL;DR
- Hackers are increasingly using AI tools like WormGPT to generate phishing emails, write malware, and scale cyberattacks with minimal effort.
- Kenya has seen a rise in AI-enabled attacks, even as overall cyber threat reports have dropped, with government systems like e-Citizen already targeted.
- AI makes impersonation attacks easier, allowing criminals to clone voices and craft personalized scams that are harder to detect.
- To stay safe, you should update your systems, use strong credentials with two-factor authentication, train teams to spot suspicious activity, and avoid unverified AI tools.
You’ve probably heard a lot about AI lately, how it’s changing how we work, search, code, and communicate. But while companies rush to plug large language models into everything, cybercriminals are doing the same. Quietly, efficiently, and dangerously.
What used to take hours of writing malicious code or crafting phishing emails now takes seconds with the help of AI. And it’s not just a few rogue actors tinkering in the shadows anymore.
There’s an entire underground economy forming around AI-powered cybercrime, complete with custom-built AI models designed specifically to hack, phish, and defraud. These tools don’t just help hackers move faster. They help them scale.
And yes, it’s happening here too. Kenyan organizations, from banks to government portals, are feeling the effects. Even as overall cyber threats dropped in 2024, AI-enabled attacks in Kenya have increased.
Criminals are using machine learning to slip past defenses, automate their social engineering, and exploit weak systems at scale.
This is no longer a theory or a future threat. It’s already here.
How AI Is Fueling a New Wave of Cybercrime
A 2025 report from Check Point Software Technologies paints a clear picture: AI is being weaponized. Hackers are no longer just experimenting with generative models, they’re building with them, customizing them, and selling them as tools for hire.
Take WormGPT as an example, it’s a version of ChatGPT that’s been stripped of ethical boundaries. It can write phishing emails, develop malicious code, and generate fake conversations to trick people into giving up sensitive information.
There are others too, GhostGPT, FraudGPT, and HackerGPT, each designed to serve specific parts of the cybercrime process.
Some of these are jailbroken versions of mainstream models. Others are custom-trained from open-source projects like DeepSeek or Alibaba’s Qwen, picked specifically because they have few restrictions and can be run locally or anonymously. Cybercriminals are even packaging these tools with subscriptions, tutorials, and Telegram channels for support.

But it’s not just about models. Entire fake platforms posing as AI tools have popped up, offering downloads that are actually malware in disguise. One example is HackerGPT Lite, which masquerades as a lightweight AI assistant but is suspected to be a phishing site.
Another case involved a Chrome extension pretending to be ChatGPT, which hijacked Facebook sessions to steal login details. These tools lower the barrier to entry for cybercrime. Even low-skill actors can now launch high-level attacks with help from AI.
The Kenyan Threat is Getting Smarter
In October 2024, the Communications Authority of Kenya (CA) issued a warning: while the total number of reported threats dropped by 41.9% in the previous quarter, the number of AI-enabled attacks was rising.
“Cybercriminals are increasingly using AI-enabled attacks to enhance the efficiency and magnitude of their operations,” said CA Director-General David Mugonyi.
He pointed to attackers automating phishing campaigns, exploiting open ports, and using AI to probe for system misconfigurations and weak access controls.
Also Read: Cyber threats in Kenya skyrocket, with 2.5 billion attacks recorded in three months
Between July and September 2024, Kenya’s National KE-CIRT/CC recorded over 650 million cyber threat events. Many were driven by botnets and automated tools powered by machine learning. AI is helping criminals run persistent, coordinated attacks at scale, fast enough that most systems can’t keep up.
A major case that rattled Kenya was the e-Citizen attack, attributed to Anonymous Sudan. Government services were knocked offline for days.
While the full extent of AI involvement in that attack wasn’t confirmed, experts believe automated vulnerability scanning and AI-generated payloads were likely used to breach defenses.
Impersonation is Getting Easier and Harder to Detect
One of the more dangerous applications of AI is impersonation. With deepfake tools and natural language models, attackers can create voice clones of executives or write emails that sound like they came from your boss.
And these aren’t generic scams, they’re hyper-personalized. AI helps attackers mine public information to tailor messages that feel legitimate.
In some cases, victims have received WhatsApp voice notes from what sounds like a known contact, asking them to transfer money or share credentials. These aren’t just phishing attempts—they’re AI-assisted fraud in real time.
What You Should Do
If you’re online, you’re a target and that’s the reality. But there are a few practical steps you can take to make things harder for attackers.
Start by keeping your software and systems up to date. Many breaches happen because of known vulnerabilities that were never patched. It sounds simple, but it’s one of the most effective defenses.
Next, make sure your admin accounts and other high-level access points use strong passwords and two-factor authentication. Attackers often rely on weak credentials or poorly secured endpoints to get in.
Train your staff or anyone you work with to recognize modern phishing attempts. AI-generated scams can be slick, grammatically perfect, well-written, and often personalized but they still leave clues. A sudden request for payment, odd timing, or unexpected urgency should always raise flags.
And be skeptical of new tools, especially those offering “free” AI services or downloads. Some are just malware in disguise. Don’t install browser extensions or applications unless you’ve verified they’re legit and come from a trusted source.
Being cautious doesn’t mean being paranoid. It just means staying alert to how the game is changing—and not making it easy for someone on the other end of an AI tool.





Leave a Reply