NSW announces framework to strengthen AI oversight

The NSW Government has announced a new Artificial Intelligence Assessment Framework to give its agencies clearer guidance, stronger safeguards and more efficient assessments of AI systems.

Developed by the recently launched NSW Office for AI, the new framework is designed to make it easier for agencies to identify, manage and mitigate risks to ensure the use of AI is safe, fair, inclusive and transparent.

Using the foundations and principles of the previous framework, the Office for AI has redesigned how this vital tool looks, feels and works to ensure it is aligned with national and international standards including the Commonwealth National Framework for the Assurance of AI in Government and the European Union AI Act.

The new framework replaces lengthy, subjective self-assessments with a faster, standards-aligned approach that automatically identifies the right level of oversight and expert review. The Office for AI says that it reduces assessment time from days to less than 30 minutes with low-risk systems able to move through the process quickly, while higher-risk or critical systems are automatically identified for review.

This ensures agencies identify risks early and apply the right governance and mitigations to maintain community confidence in the safe and responsible use of AI.

For example, a public-facing AI chatbot, which helps people apply for government services, would be assessed as high risk. This is because the chatbot could collect sensitive personal information and can influence the advice people receive, so mandatory safeguards are triggered, such as a Privacy Impact Assessment, cybersecurity and legal review, and accessibility requirements. It would also be submitted to the AI Review Committee for independent oversight.

Ethical principles are embedded in the framework’s logic, helping agencies identify when additional safeguards are needed, such as bias testing, accessibility checks and human-rights screening, meaning appropriate safeguards can be in place before deployment.

The redesign was delivered in partnership with CSIRO’s Data61, and forms part of a broader program to modernise digital assurance in NSW. This work is supported by additional whole-of-government guidance, including on the use of emerging technologies such as agentic AI.

“AI can transform government services, but we have a responsibility to use it safely. This framework ensures agencies identify risks early and apply the right safeguards,” said Minister for Customer Service and Digital Government Jihad Dib. “As AI starts to be embedded in projects as business as usual, it is essential we have clear guidelines for its implementation and ensure its use is transparent, inclusive and fair.”

“AI is rapidly transforming how NSW Government agencies work and deliver services, improving productivity and supporting economic growth while putting people first,” said the Head of the Office for AI, Daniel Roelink. “The redesigned framework is simpler and more accessible for all public servants, informed by process engineering from CSIRO and extensive testing across multiple agencies.”

More information on the AI Assessment Framework can be found here.

Image credit: iStock.com/Natee127