evidoria

← Back to browse

Good practice Imported

Australia's AI Transparency Statements — a Mandatory Register Built After an Independent Audit Found Compliance Gaps

Australia · Canberra · See the Australia profile · See the Canberra profile

Evidence: Observational / pre–post Top 6% 87/100 · Ask Evidence Copilot about this practice

Canberra required 94 federal agencies to publish AI-use transparency statements by 28 Feb 2025. The Digital Transformation Agency says all complied — but independent research across 224 agencies found only 29 statements were easily discoverable, prompting a new central register.

94
Agencies mandated to publish transparency statements
114
Total transparency statements published (mandatory + voluntary)
20
Agencies publishing voluntarily
224
Agencies audited by the ADM+S Centre
29 of 224 agencies audited
Statements easily discoverable from an agency's own website
101 of 224 agencies audited
Statements found via a deeper search

Details

Maturity
Established
Promoter
Digital Transformation Agency (DTA), Australian Government
Period
2024–2025 (mandatory deadline 28 Feb 2025; central register launched 2025)
Keywords
algorithmic transparency, digital government, AI governance, public administration

Context

Under Australia's Policy for the Responsible Use of AI in Government, the Digital Transformation Agency (DTA) required non-corporate Commonwealth entities (excluding Defence and intelligence agencies) to publish a public AI-use transparency statement by 28 February 2025.

Activities

The DTA reports that all 94 mandated agencies met the deadline, joined by 20 more publishing voluntarily for 114 statements in total, required to be updated at least annually or whenever AI use changes materially; independent researchers at the ADM+S Centre, working with the Office of the Australian Information Commissioner, then audited 224 federal agencies using automated site crawling, targeted search and manual inspection.

Results

The independent audit found only 29 statements were easily discoverable from an agency's own website, with a deeper search turning up 101 in total, and noted that most were not linked from agency homepages and that no penalty existed for non-compliance; in response, the DTA built a centralised public register of AI transparency statements to address the discoverability gap.

Conclusions

The episode illustrates a real feedback loop between independent academic oversight and a policy fix, set against a backdrop of low public trust in government AI in Australia following the Robodebt royal commission's findings on inadequate algorithmic safeguards.

Implementation

Indicative cost
Medium (€50k–€500k)
Time to results
Medium (1–3 years)
Staffing & skills
The Digital Transformation Agency (DTA) administers the mandate and the central register; the independent audit was conducted by the ADM+S Centre with the Office of the Australian Information Commissioner.

Conditions for success

  • An independent academic audit exposing a discoverability gap fed directly into a government policy fix (the central register).

Common failure modes

  • No penalty existed for non-compliance with the transparency-statement requirement.
  • Most agency statements were not linked from agency homepages, undermining public discoverability.

Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.

Data sources

Where this practice's information was retrieved from, and when.

Similar practices you may find useful