Government agencies fail first hurdle under AI self-reporting policy

According to data released by Australia's Digital Transformation Agency, only 87 of 200 federal agencies submitted their reports on time in the first assessment period of the artificial intelligence transparency policy. According to ABC News Australia, even large agencies such as the Departments of Health and Social Services were late in their submissions. The policy requires agencies to document publicly which systems use artificial intelligence, which datasets are used in training, and their risk assessments.

Compliance Lead Sandra McDermott said the vast majority of agencies did not maintain an artificial intelligence inventory and struggled to interpret the reporting format. Helen Sullivan, professor of public administration at the Australian National University, said the policy relying on voluntary self-reporting was a structural weakness. Government Services Minister Pat Conroy said the additional time was a one-off and that agencies must fully comply in the coming quarter.

The policy was rolled out as part of the Albanese government's 2025 artificial intelligence regulatory framework. Civil society organisations have called for legal obligations to replace voluntary reporting. Business representatives meanwhile said they were concerned that the corporate sector might be drawn into similar reporting requirements.