Security News

ACS Warns Australia's AI Safety Plan Needs Skilled Local Experts to Work

Key Takeaway If your business uses AI tools, name a responsible owner who understands them and has the authority to pause or stop their use if something goes wrong.

The Australian Computer Society (ACS) has warned that the Federal Government's AI safety package will not succeed without capable, trusted technology professionals to put it into practice. Responding to the package announced yesterday, ACS argued that basic AI literacy across the public sector is not enough to protect public infrastructure. It said that if an AI agent gains unauthorised access to a government system, specialist, locally resident expertise is needed to work out what data or systems were accessed, assess wider risks to public services and act quickly. The same skills are needed to independently test powerful AI models and check the safety claims made by developers.

ACS pointed to the "Rapid Review into the OpenAI incident involving Medicare". It argued that the Government's six proposed protections and the review's 18 recommendations do not go far enough on skilled people authorised to investigate and escalate AI risks inside the public service. ACS CEO Dr Prins Ralston asked: "Where are those people in the plan? How will we know they have the capability to do the job?" ACS also cited research indicating that only 23% of Australians trust AI developers to ensure their software is safe.

ACS called for clearer capability and accountability rules for roles overseeing high-impact AI in government. It suggested assessing those roles against recognised skills frameworks such as SFIA, and requiring named accountability for State-used or State-funded AI: a senior owner, a capable resident technology professional, and an officer authorised to pause or withdraw systems. It also proposed that offshore developers training covered large-scale models appoint a qualified compliance officer in Australia, and that the public service build specialist AI and cyber skills.

AI safety Australian Government ACS cyber capability AI governance
Building or buying AI systems? Governing them under ISO 42001 ->

Summarised by CISO AI from Australian Cyber Security Magazine, written with Claude Sonnet 5.5. We link back to every original so you can read it yourself.