NSW government AI projects face ethics assessment under new assurance framework


All New South Wales government agencies using AI will be required to meet best practice ethical requirements under the country's first mandated AI Assurance Framework, which comes into effect today. The framework, which was developed by the NSW Advisory Committee led by NSW chief data scientist Ian Oppermann, has been designed to ensure AI-based government projects are safe, ethical, and can be integrated with future technologies. It also assists agencies with risk mitigation strategies and establish clear governance and accountability measures. "From diagnosing sepsis in hospital patients to identifying drivers illegally using mobile phones while driving, the NSW government is already using AI to improve the lives of NSW residents," Oppermann said. "As the technology evolves and becomes more sophisticated, the Framework will ensure projects remain transparent and include the highest levels of privacy, security and assurance, so customers can feel even more confident when dealing with the NSW government. "Mandating the framework will ensure all NSW government services using AI are required to implement strong privacy and data management safeguards." Projects with budgets of more than AU$5 million or supported by the Digital Restart Fund will also be subject to assessment by the AI Review Committee to ensure compliance under the new mandate. The state government added the only exceptions where the AI Assurance Framework will not apply is when a project uses an AI system that is a widely available commercial application, and the solution is not being customised in any way or being used other than intended. The framework is part of the state government's AI strategy in which it has pledged that transparency will be the focus and vowed to make the state the digital capital of the southern hemisphere in the next three years. "AI stands for absolutely imperative for the new New South Wales.

Duplicate Docs Excel Report

None found

Similar Docs  Excel Report  more

None found