The principle "A compilation of existing AI ethical principles (Annex A)" has mentioned the topic "safety" in the following places:

    11. Robustness and Security

    AI systems should be safe and secure, not vulnerable to tampering or compromising the data they are trained on.