SA Enterprises Race to Deploy Agentic AI, But Security Gaps Remain

Summary: South African enterprises are rapidly deploying agentic AI that can plan, act, and make decisions autonomously, but most lack corresponding security foundations. Globally, three out of four agentic AI projects expose organizations to major security risks, with the local market projected to reach USD 152 million by 2030. Key Points: The shift from 'AI that talks' to 'AI that acts' introduces new risks, such as unmanaged agentic identities and unpredictable runtime behaviors. Traditional security tools are not designed for a world where software agents hold identities and act independently. Challenges include limited visibility across agent deployments, difficulty assessing evolving risks, and threats like shadow AI agents, prompt injection attacks, and agent hijacking at the user layer. Analysis: Organizations must secure agentic AI at both infrastructure and browser layers to scale AI-powered enterprises confidently. Those addressing security early will scale faster, while others risk spending more time on cleanup than progress.