# Dozens of Australian government agencies failed first test at self-policing AI use

Canonical source: https://webcam.substack.com/p/dozens-of-australian-government-agencies
Publication: webcam
Published: 2026-06-12T06:27:17+00:00
Content role: commentary

## Summary

The early stumbles shows the challenge of leaving AI rule-making to each department by themselves

## Public text

Hi!

I just published: Government agencies fail first hurdle under AI self-reporting policy

A Friday treat off the back of the Four Corners report earlier this week about Australia’s approach to rules around AI.

After last year’s election, the federal government did a U-turn and decided against EU style regulation with a central AI Act and broad rules for its use.

Instead, it tasked all the government department and agencies with setting their own rules for AI.

But around the same time, many of these agencies showed the flaws of this approach.

Dozens of them failed at their first test of self-policing their own use by missing deadlines for disclosing how they use AI or who is responsible.

While they’ve remedied this issue, it’s notable because the shows the potential issues with adding another issue to the already full plates of these government agencies — particularly for an issue that’s as complicated and fast moving as AI!

Thanks, CW

P.S. I love tips . I also love hearing what people think about stories and can promise 100% confidentiality for anyone who contacts me. Message me on encrypted messaging app Signal or send me a secure email here .