Business
Agentic AI Is Changing the Rules of Identity Security in South Africa

Agentic AI is rapidly moving from a tech trend to an everyday business reality in South Africa, and it is reshaping how companies think about security.
Unlike generative AI, which works within prompts and predefined rules, agentic AI can act, learn, and adapt without constant human oversight. This autonomy makes it a powerful tool for efficiency and scalability, but it also creates urgent new challenges for identity and access management.
Why AI Needs Human-Level Access
To do their jobs, AI agents often require the same privileges as human employees. They can access resources, interact with other systems, and optimise operations without waiting for approval. That convenience also gives malicious actors more opportunities if security controls are weak.
Traditional identity and access management (IAM) systems were never designed for independent, non-human “workers.” This gap can be exploited by hostile AI agents, which could learn to bypass controls, escalate privileges, and compromise sensitive systems.
A Workforce Approach to AI Security
Cybersecurity experts recommend treating AI agents as part of the workforce. That means structured onboarding, changing access rights as roles evolve, and granting permissions only when needed through “just-in-time” access.
Just like human employees, AI agents need constant monitoring. Tamper-proof logs, cryptographic signatures, and transparent audit trails help track every action and quickly detect any suspicious behaviour.
Tools for the New AI Era
Identity security provider Delinea offers solutions that embed policy enforcement and access controls throughout an AI agent’s lifecycle. This ensures agents have the resources to work effectively while limiting risk.
Through local distributor Altron Arrow, South African businesses can implement Delinea and other advanced tools to adopt agentic AI safely and responsibly.
The Cost of Standing Still
With cybercrime already costing the country billions each year, identity security is no longer an IT back-office concern; it is a business-critical issue. Agentic AI can be a competitive advantage, but without modernised safeguards, it can also be a fast track to reputational and operational damage.
The message for South African companies is clear: securing identity in the age of agentic AI is not optional; it is essential.
Also read: Sim Tshabalala’s Exit Marks a Changing of the Guard at Standard Bank
Follow Joburg ETC on Facebook, Twitter, TikTok and Instagram
For more News in Johannesburg, visit joburgetc.com
Source: MyBroadband
Featured Image: TRENDS Research & Advisory