Investigation Uncovers Hundreds of AI Use Cases Across Federal Departments

A series of freedom of information disclosures has pulled back the curtain on artificial intelligence adoption across the Australian public sector, revealing hundreds of active use cases across at least 20 federal departments and agencies.

The findings show widespread experimentation with automated tools to draft documents, summarise large bodies of evidence, and analyse internal workflows. Beyond standard generative chatbots, several agencies are trialling sophisticated autonomous agents capable of orchestrating complex tasks across multiple systems and data sources.

Experimental agents in key agencies

The documents show AI deployment spanning essential bodies, including the Australian Securities and Investments Commission, the Australian Competition and Consumer Commission, the Department of Health, the Department of Agriculture, and the Commonwealth Ombudsman.

Among the disclosures, the Department of Industry, Science and Resources logged 19 distinct use cases. These involve both custom-built and off-the-shelf agents designed to run analysis, produce drafts, and coordinate actions across different digital platforms. While the department maintained that human reviewers check outputs to catch errors, omission, and bias, the breadth of deployment highlights just how deeply automation is taking root in core administrative machinery.

Governance advocates and technology analysts have voiced concern that while departments must publish broad transparency statements, registers detailing specific deployments remain largely shielded from public scrutiny. Experts also cautioned that without dedicated bodies actively monitoring unapproved or informal tool adoption, unvetted use cases could quietly steer public decisions.

What this means for Australian businesses

The deep footprint of AI across government bodies provides valuable lessons for small and medium enterprises managing their own digital transformation:

• Regulators are using the tech themselves: With watchdogs like ASIC and the ACCC already building internal AI literacy, small businesses should expect regulators to become faster and more sophisticated at analysing compliance data and market conduct.
• Keep detailed use-case registers: Government departments are coming under scrutiny because they failed to document and disclose automated systems clearly. Small businesses should maintain a simple internal register tracking which AI tools staff use, what data feeds into them, and who signs off on the results.
• Human-in-the-loop remains non-negotiable: Autonomous agents can coordinate multistep tasks, but they still produce hallucinations and biased summaries. Retain explicit human checkpoints for customer-facing advice, quoting, and financial decisions.
• Guard against shadow IT: Staff often adopt consumer generative tools without formal approval to save time. Provide clear company guidelines on permitted software so confidential company and client information never leaks into public systems.