Security must be rethought in the age of AI

The central message of a much-noticed keynote speech put it in a nutshell: the IT world is in a structural imbalance. While on the one hand there is an attack industry worth billions - with business models such as „attack as a service“, „hacking as a service“ or automated attack waves - on the other hand there are often fragmented, underfunded defense structures. The use of AI has dramatically increased the threat level - not only through AI-based attacks, but also through the use of AI. In data poisoning, for example, attackers falsify the AI model during the training phase and build in hidden vulnerabilities or backdoors.

Adversarial evasion aims to deceive an agent's logic by making barely perceptible changes to its input so that data is misclassified. In prompt and tool injection, agents are induced to use authorized tools for unauthorized, malicious actions.

Integrative approach to threats

This development has direct consequences for companies. Cyber attacks are not only becoming more frequent, but also more precise, more scalable and more difficult to detect. Traditional security mechanisms are quickly reaching their limits. One reason for this is that IT security is still „tacked on“ to many projects, i.e. only considered at the end. In view of the current threat situation, this approach is no longer expedient. What is needed is an integrative approach in which security is part of every architecture and development decision from the outset. This applies not only to software development, for example controlled software supply chains, but above all to operations. This is where a dangerous gap becomes apparent in practice: While prototypes and proof-of-concepts are often implemented quickly, there is a lack of well thought-out security concepts in the transition to productive use. This is particularly true in the SAP environment. Traditionally, many SAP systems were considered to be relatively isolated - a kind of „secure castle“ that was protected by its isolation. But those days are over. With strategies such as cloud transformation, hybrid landscapes and side-by-side integration, these systems are increasingly opening up to the outside world. Interfaces to third-party applications, APIs and cloud services are creating new dependencies - and therefore new attack vectors.

Open source in the security context

Against this backdrop, the use of open source technologies is becoming increasingly important in the security context. Their benefits are not limited to development environments - there is also a broad spectrum of established frameworks and tools in operation. Red Hat OpenShift offers more than 20 frameworks for the areas of Control and Protect as well as Detect and Respond, which cover the entire DevSecOps lifecycle. These include solutions such as „Application Configuration Analysis“, „Risk Profiling“ or „Runtime Behavioral Analysis“ and „Auto-Suggest Network Policies“. They demonstrate how modern security mechanisms can support ongoing operations today.

The discussions at the Technology Days clearly show that IT security is no longer a marginal issue, but a key success factor for digital transformation. Companies are required to adapt their strategies - both technologically and organizationally. This also means not delegating responsibility to individual providers or products alone. Instead of relying on supposedly secure „black box“ solutions, companies should focus more on transparent, comprehensible and integrable security concepts. Open source can play a decisive role here - as a flexible and powerful building block within a holistic security strategy.

To the partner entry: