How the Rise of AI is changing Cyber-security
For years, cyber-security operated on the assumption that cyber-attacks depended heavily on human intelligence and expertise. Scammers relied on phishing emails, but these were often generic and easy to spot by a careful reader. Today, that premise no longer applies, as artificial intelligence (AI) has changed the cyber-security landscape.
As one of the the most recent and powerful technologies, AI allows computers and machines to perform tasks that would normally require human intelligence. Undoubtedly, AI when and if used well can save time and help people work more efficiently. Unfortunately, criminals are now also using it to plan and carry out cyber-attacks, a reality that has added to the broader cyber space threats facing Kenya and other parts of the world.
For instance, between July 2025 and June 2026, the National Kenya Computer Incident Response Team recorded 11.12 billion (B) attempts to attack computer systems, which was 29 per cent more than the 8.62 B attacks recorded the previous year. From April to June 2026, more than 2.25 B of these attempts targeted computer systems. The team also recorded nearly 59 million (M) cases of harmful software and over 24 M attempts to break into accounts through repeatedly guessed passwords. Though many of the attempts failed, the huge volumes highlight a growing threat.
Previously, carrying out attacks on this scale would have required coordinated groups to spend weeks gathering information and preparing each attempt. Today, with the use of AI, criminals can target computer systems and accounts more quickly and at a greater scale. This is possible because AI can produce highly convincing emails, imitate voices, clone images and videos, and personalise messages within minutes. As a result, sophisticated cyber-attacks that once demanded specialized expertise are now executable through accessible AI tools.
For banks, this changing threat landscape calls for a more robust approach to cyber-security. As AI makes impersonation and social engineering more convincing, familiar signals, such as an email that appears to come from a senior executive or a voice on a call that sounds like a trusted colleague, can no longer be treated as proof of identity. Banks must strengthen verification across digital and human interactions, monitor transactions for unusual patterns and equip employees and customers to recognize emerging scams.
Yet, AI-driven systems learn what normal activity looks like and can flag subtle deviations, such as unusual login locations, unexpected transaction sequences, or attempts to access restricted information. By analysing large volumes of data in real time, these tools can help security teams identify potential threats earlier and focus their attention on activity that warrants investigation.
Beyond detecting suspicious activity, AI can also help banks test their defenses before attackers find a way in. It can check software and systems for weaknesses and explore how those weaknesses might be exploited. Teams can then prioritize fixes and strengthen vulnerable areas. By revealing weaknesses early and helping teams to focus on the most urgent fixes, AI-assisted testing allows banks to strengthen their systems before those gaps can be exploited.
These measures are particularly important because individuals are increasingly exposed to AI-enabled impersonation. Using AI, fraudsters can personalize their messages to individuals using publicly available information and impersonate contacts, making it more difficult to distinguish between genuine and fraudulent requests, particularly when the communication creates a sense of urgency.
Evidently, the question is no longer whether an individual, or organization, can recognize a sophisticated scam; It is whether the system’s safeguards and response mechanisms in place can prevent an attempted attack from becoming a costly breach.
Kenya’s cyber incidents figures illustrate the challenge. Billions of attempts create a persistent threat that cannot be addressed through awareness alone. Individuals and institutions must combine vigilance with practical safeguard.
For customers, that means setting up safety nets before you need them. The best defense includes using a different password for every account, switching on two-step verification, and activating transaction alerts to identify suspicious payments within minutes. For banks and businesses, it means assuming some attempts will succeed and designing for that by strengthening controls against unusual transfers, detecting suspicious behavior early, and, as well, ensuring that customers can report suspected fraud and get assistance quickly.
Given that reality that AI has made attacks cheaper and easier to launch, the best response is not only to make attacks harder to execute, but also to make successful breaches more difficult to achieve and less costly to recover from.
By Michael Etale
The writer is Chief Information Security Officer (CISO) at Absa Bank Kenya
