The term General Adversarial Robustness is primarily found in the fields of Artificial Intelligence, Cybercrime and Cybersecurity, and Digital Transformation. It describes how resilient an AI system is against so-called „adversarial attacks“. These are tricks or manipulations used by hackers to fool or deceive an AI.
Imagine an AI monitoring surveillance cameras and automatically identifying suspicious individuals. An attacker might try to trick the AI by making small changes to their appearance – for instance, through specific patterns on their clothing – so that the AI doesn't recognise them. General adversarial robustness means that the AI continues to work reliably even if someone attempts such attacks.
Another example: In self-driving cars, traffic signs are recognised by an AI. If someone sticks stickers onto a stop sign, a weak AI might no longer correctly recognise the sign as a stop sign. In contrast, a system with high general adversarial robustness remains safe and continues to make the correct decisions.
Anyone betting on artificial intelligence should therefore always check how „robust“ their system is against such targeted deception attempts.
Deutsch 
English (UK)