TNG Technology Consulting GmbH

Insights from BTD26: TNG on AI safety and the importance of red teaming

November 18th, 2026

AI models and AI-enabled systems can be deliberately attacked, making systematic security evaluation an essential part of their development. At Big Techday 26, Aron Erdelyi, Benjamin Balogh, and Laszlo Balazsik discussed today's most relevant attack vectors as well as defensive techniques.

Our colleagues explained how attacks such as prompt injection and jailbreaking can bypass safety mechanisms. They also showed why red teaming and structured evaluation are key to identifying vulnerabilities early and improving the robustness of AI systems.

Check out the recording here.