The AI Strategy sets the direction on the development and use of AI by NSW Government agencies and is underpinned by the mandatory AI Ethics Policy and AI Assessment Framework. The aim of the NSW Government’s approach is to support agencies to innovate with AI technology, while making sure it is used safely and responsibly, with clear accountability for the design and use of AI Systems.

The NSW Digital Assurance Framework is the all-of-government assurance function for digital solutions, including AI in NSW. Any agency project over $5 million or funded by the DRF will be assessed to determine if they need to apply the AIAF, and if so, they will be directed to do so. This control is being implemented into the NSW Digital Assurance Framework.

Agencies are required to adhere to the following mandates in their development, procurement, and use of AI systems:

1. The AI Ethics Policy

The AI Ethics Policy provides a principles-based approach to applying AI in NSW Government. The Policy outlines five key principles that will ensure that AI is only used where there is a strong case for doing so, generating safe and responsible outcomes.

Applying the AI Ethics principles of Community Benefit, Fairness, Privacy & Security, Transparency, and Accountability ensures that any use of AI systems meets the highest ethical and assurance standards and that proactive risk management strategies are in place.

2. The AI Assessment Framework

Commencing on 31 March 2022, the AI Assessment Framework, formerly the AI Assurance Framework, provides practical guidance for designing, building, and using AI technology. The Framework supports agencies in assessing risks, implement mitigation strategies, and helps establish clear governance and accountability measures. The Framework also requires agencies to detail how their projects align with the five ethical principles contained in the AI Ethics Policy.

The updated AIAF associated with this circular now includes additional guidance to ensure NSW has a standard and consistent approach to AI risk levels, generative AI, and contemporary best practices to reduce the likelihood of harm.

Agencies are required to utilise the AI Assessment Framework throughout all stages of their AI system lifecycle, from inception to retirement, and to conduct periodic reviews based on the potential risk of harm as determined by the AIAF.

3. High-risk or higher investment projects must be submitted to the AI Expert Review Committee

After applying the AIAF and implementing all risk mitigations, any residual high and very high-risk projects/systems must be submitted to the AI Review Committee. This committee, comprising industry experts from all relevant domains, will provide additional recommendations to strengthen the management of high-risk AI systems.

There are two ways to engage the AI Review Committee.

1. Guidance will be provided via the NSW Digital Assurance Framework when you register your project for: Projects >$5m and Digital Restart Funded projects.

2. Direct via emailing the [email protected], for high risk projects: <$5m and operational/ongoing systems.

Note:

No amendments were made to either the AI Ethics Policy or the AI Strategy. Both were last updated in March 2022.