AI changes the speed and scale of cyber risks, but not the fundamentals

For all the uncertainty surrounding artificial intelligence (AI), one thing remains certain – its impact on the security landscape.

Where enterprises are concerned, this is true on two very important fronts, writes Chris Mulder, senior solutions architect at Red Hat.

The first is traditional cyber threats such as scams and fraud. This is where threat actors leverage the tech to target multiple victims and systems, making it easier to disguise fraudulent interactions, and accelerate the pace of their attacks. The impact is felt across South Africa, which is now the centre for cybercriminal activity on the continent, recording significant levels of ransomware and phishing detections and DDoS attacks.

The second front lies in software security. According to research by Google Threat Intelligence Group, AI is not only accelerating the pace of vulnerability discovery and exploitation, but also changing the types and risk profiles of said vulnerabilities. Organisations are also concerned with the exposure of their own AI systems, wondering whether their security protocols are up to scratch or whether their deployed models are both secure and ‘well behaved’.

Local businesses now have to increase their security resilience, but to what end? A cybersecurity posture is not one taken after the fact; it is operationalised from the very beginning. Businesses achieve this by sticking to the fundamentals and aligning them with the nature and scale of their infrastructure.

Fundamental 1: Secure the platform, not the model
A big benefit of the AI revolution during the last five years is that it has forced organisations to consider the performance and capabilities of their underlying IT infrastructure.

Many South Africans will prosper with the help of consumer-facing, off-the-shelf solutions. But those looking to develop and deploy their own models and applications must modernise their foundations, become familiar with automated workflows and consider open and flexible platforms.

That also speaks to a fundamental idea about enterprise IT: you cannot build a secure application with an insecure foundation. Before a team even reaches the point of considering what models to use (i.e. frontier LLMs or otherwise), the platform represents core infrastructure through which data flows, operations are executed, and access and overall security are enforced.

With the right platforms in place, organisations can use tools and resources consistently to execute models and deploy agents. They also forge the security and governance policies to protect against continuous cyber risks. As more tried, tested and validated models become available for integration and inference, organisations can integrate with confidence and at the scale they need to.

**Fundamental 2: Check your sources
**Many businesses assume that, when it comes to software development, it’s an eternal struggle between open source and proprietary. The truth is that modern software development runs on open source. It’s innovation driven by the power of community. But with that truth comes the risk of unverifiable provenance, the idea that organisations don’t know where software comes from, what went into it, and what vulnerabilities it may have.

Enterprise-grade platform vendors help organisations address this risk with the help of trusted open source libraries that offer full software supply chain transparency. These libraries provide organisations not only offer upstream monitoring, but also service-level guarantees that form the basis for trustworthy application development.

When it comes to software, organisations will always have to deal with dependencies. The same goes for vulnerabilities, which precipitate the need for security-centric development practices. However, by knowing where software comes from (something that also applies to AI models), they can establish a level of trust and have confidence in their technology ecosystems.

**Fundamental 3: Autonomous security is possible and beneficial **
As AI identifies (and potentially exploits) larger volumes of vulnerabilities, businesses need to take their security posture to a new level. At the same time, AI introduces higher levels of complexity. Distributed architectures and more data-driven applications mean businesses have wider and more complex environments to oversee.

This is where IT automation comes in. Today, organisations need to deliver patches, streamline compliance, and establish perimeters and guardrails with greater speed and efficiency. Using a trusted automation platform, organisations can execute rapid patching during scheduled maintenance periods, be alerted to newly identified vulnerabilities, isolate and contain the resulting risks, and potentially remediate issues. And they do this all while remaining compliant and maintaining an audit log of every action taken. Automation also opens the door to advanced AIOps. Organisations can manage issues in real time and use AI-driven guidance.

Critically, automation also helps answer the scale question. It lets organisations maintain the speed of issue remediation and their consistency in doing so, regardless of system size or complexity. Even implemented for the most basic of processes, automation gives organisations a leg up as they build out their AI infrastructure.

Fundamental 4: Get the right human for the job
While automation represents the next frontier in enterprise IT security, it is not without its shortcomings. Nor is it a single, end-all solution for organisations. While AI can support teams in finding and patching vulnerabilities, it cannot provide a risk-based security strategy. It also cannot, on its own, deliver the integrity that keeps organisations upright at a time of crisis. That is still the responsibility of the IT team and comes down to the skills they wield.

South Africa’s tech jobs market is currently on the rebound as roles, especially data-focused ones like data scientists and data warehousing, are needed as businesses move to the cloud and explore AI’s potential. This is notable because there was a point not long ago when many business leaders saw AI as a means to lower headcounts, even in IT departments. However, the rise of agentic AI has thrown the importance of human oversight and expertise into sharp focus.

People will always be fundamental to the success of technology. By investing in the right talent and leveraging local skill sets, organisations can trust in the systems they’re building while laying the foundation for future technology implementations.

Enterprise IT will always be subject to some kind of risk, both internally and externally. As South African businesses evolve and enhance their use of technology, the risks will also evolve. But by sticking to the fundamentals, they can protect themselves and uphold their security postures. And good posture always begins with solid footing.

And with good posture, they will always walk upright.

And with each organisation that prioritises security in this way, they help strengthen our market and demonstrate what’s possible in this era of enterprise IT.

The post AI changes the speed and scale of cyber risks, but not the fundamentals appeared first on IT-Online.