African Firms Face Agentic AI Security Challenges

Africa is embracing autonomous AI at an unprecedented pace, but its security model has not kept up. According to Kerissa Varma, Microsoft's Chief Security Advisor for Africa, traditional cybersecurity strategies are breaking down as AI agents become more common in industries like finance and logistics.

Varma notes that 82% of African respondents to a survey by MIT Sloan Management Review and BCG view AI agents as colleagues rather than tools. However, this rapid adoption has left organisations struggling to navigate an increasingly complex risk landscape.

The challenge is not only that organisations are deploying autonomous systems faster than governance can evolve, but also that cyberthreats themselves are becoming more autonomous. This shift in the 'physics of cybersecurity' requires a new approach, one that involves redesigning security around a cyber stack built for autonomy.

Varma advocates for a multi-layered system comprising signals and sensors, security context, models, and actuators. Each layer must be designed to work together seamlessly, enabling AI agents to apply intelligence in real-time while maintaining strong oversight and accountability.