Engaging with Artificial Intelligence (AI)

This joint guidance from the Australian Signals Directorate (ACSC/ASD) and international partners summarizes common AI-related threats — including data poisoning, prompt injection and adversarial examples, generative AI hallucinations, privacy/IP risks, and model stealing — and provides practical mitigation considerations for organisations (secure-by-design, MFA, least privilege, backups, trials, logging/monitoring, incident response, and adherence to relevant frameworks such as NIST and MITRE).