EDITOR ’ S QUESTION

When that happens , the detection , threat hunting , forensics , and response can all be automated at machine speed with AI to counter the threat . Therefore , fighting AI with AI , it is the defence for a modern malicious offense .

Every mature cyber security organisation performs some form of threat hunting on a regular basis . That is , you assume you have been compromised , even if you have not , and look for clues to support your conclusion .

Typically , this is a laborious process of log , activity , and application reviews to find a bread crumb in the ether . AI defences can provide those initial gold flakes that indicate a compromise in support of threat hunting that are buried in all the information .

In other words , AI defences used for pattern and behavioural recognition can speed up detection time of anomalies faster that any human manually performing the process .

Certain patterns , like spray attacks and session hijacking , observed by AI , provide a strong indicator that an incident is in progress . Based on these results , AI can automate the response from forced password changes and entitlement reviews , to account disablement .

To combat this , modern cybersecurity solutions are integrating advanced AI models that analyse micromovements , pupil responses , and infrared patterns , making it harder for deepfakes to succeed .

Multi-layered approaches that combine facial recognition , voiceprints , and behavioural pattern analysis offer enhanced protection by adding complexity to authentication processes . An emerging best practice is to verify user identity not only during login but throughout active sessions , reducing the risk of account hijacking .

Another pressing challenge is the use of AI to mimic natural conversations for fraudulent purposes , particularly in support scenarios . The ease of access to powerful LLMs makes this an easy to implement and often highly effective option for cyber criminals .

To counter this , security solutions now incorporate context-aware monitoring , which cross-references interaction data with historical patterns to flag inconsistencies . Specialised tools capable of recognising AI-generated voice patterns are critical for identifying fraudulent calls .

Complementing these measures , tracking subtle behaviours , such as typing patterns and navigation , can confirm the legitimacy of users during interactions , ensuring robust security without compromising user experience .

AI has also exposed vulnerabilities in account recovery workflows . Attackers exploit static verification processes such as security questions , registered mobile numbers and so on , prompting the need for dynamic alternatives .

Leading solutions now replace traditional security questions with context-based challenges that assess user-specific knowledge , such as recent activities or behavioural patterns . By tailoring account recovery processes to individual risk profiles , organisations can thwart large-scale exploitation attempts .

Social engineering is another powerful tool in the arsenal of attackers and here too , the rise of AIpowered social engineering adds another layer of complexity . Modern tools can counter this by analysing call or message content in real time to detect fraud indicators , such as pressure tactics or inconsistent language . Organisations are also investing in interactive training programmes that educate employees and users on recognising advanced scams , fostering a culture of awareness and preparedness .

Effective cybersecurity solutions should integrate AI-driven threat intelligence , collaborating with global organisations to maintain updated repositories of known fraud patterns . Equally important is the regular testing of defences against AI-generated phishing and spoofing attacks which ensures readiness against evolving threats . p

SERGEY BELOV , DIRECTOR OF INFORMATION SECURITY , ACRONIS

www . intelligentcio . com INTELLIGENTCIO AFRICA 29