The government has released interim guidance on the use of generative AI in the public sector, but clarifies that it does not represent a binding position on the technology. The guidance, prepared by the Digital Transformation Agency (DTA) and the Department of Industry, Science and Resources (DISR), was developed through consultation with various Commonwealth agencies. It expands upon earlier beta AI guidance and allows for limited experimentation with generative AI.
While there is no official whole-of-government policy, individual agencies have made their own decisions on what is permissible. For instance, the Home Affairs agency has restricted open access to ChatGPT and grants exceptions for specific experiments on a case-by-case basis.
DTA CEO Chris Fechner stated that the interim guidance is meant to supplement agency-specific policies on generative AI, rather than replacing them. It is intended to provide guidance to staff within APS agencies until a comprehensive government position is established.
The recently unveiled interim guidance consists of two sections: an introduction and detailed guidance with four principles for the use of generative AI, along with hypothetical use cases. The key points in the guidance include formal enrollment of staff for access to generative AI platforms, layers of approval, logging any exceptions, and seeking commercial arrangements for generative AI solutions when feasible.
The guidance emphasizes using generative AI in scenarios with low risks of negative impact, disclosing the technology’s involvement in any activity, and critically evaluating its outputs. The primary use case allowed for generative AI is as a search engine to find generic templates, such as reports or project plans, without customizing them within the generative AI tool.
