practitioner ai au-compliance frameworks
Agentic AI harnesses
ASD’s name for everything around the LLM — tools, memory, permissions. Prompt injection is not a model patch. Control what the harness can reach and do.
Glossary
AllIRacsc-glossaryaiattack-pathau-compliancecloudconceptscryptodetectiondnsextortionframeworkshardeningidentityinitial-accessloggingmalwarenetworkphishingpractitionerprivacyransomwareresiliencesocisupply-chainthreat-intelvulnerabilitywebwindows
4 entries
practitioner au-compliance privacy ai
Automated decisions in your privacy policy (APP 1.7–1.9): what must be disclosed from 10 December 2026
From 10 December 2026, an organisation covered by the Privacy Act that lets software, including AI, make or materially shape decisions that significantly affect people must say so in its APP privacy policy. The three-part test, why human review does not get you out of it, what the policy has to list, and a six-week readiness checklist.
practitioner au-compliance privacy ai supply-chain
Buying and using AI products under the Privacy Act: the OAIC's due-diligence and APP checklist
The OAIC says the Privacy Act applies to every use of AI that involves personal information, covering both what goes into a tool and what comes out. What to check before you buy, when a prompt becomes an APP 6 disclosure, why AI output can be a collection under APP 3, and the security and accuracy steps the OAIC expects.
practitioner ai hardening frameworks
Prompt injection
Instructions and data share one channel. ASD and NCSC: no fully reliable model-level fix. Put the controls in the harness.
Definitions informed by ASD's ACSC glossary. cyber.gov.au glossary