This comment includes the core issues that concern me.

I struggle with the assertion that the data is used for government decisions that benefit me. I'm sure they are to an extent, but the extent to which the data is used in this way is infinitessimal to what's implied in the census marketing IMO. For example, a decision about funding a local school or hospital will be based on data from local schools and hospitals, not from a census. A new school is not going to be built in my area because I indicate that I have 2 kids in my home.

I don't really "trust" anyone with my sensitive data but I trust government departments less than most. It's likely that at some point Australia will have a PM elected on a populist anti-immigration platform. I believe that the likelihood that census data will be weaponised against migrants is a near certainty. I understand that there are legislative protections against this, but we've seen time and again that these protections are ineffective both at home and abroad.

Re-idenfication of de-identified data is only one of many concerns, but you do not need to be "unique enough to be obvious" to be re-identified. It's a bit like fingerprinting on the web. With a handful of datapoints you can filter a set of many million down to one or two. You don't really start with a list of names and the census data and merge the two together. It's more like, you start with someone's tax return and the census data, and then you discover a whole lot more about them.

In summary, the benefits of completing the census just don't seem to stack up against the risks. Perhaps they did 40 years ago when it was a handful of questions about language ability and earnings capacity, but not in 2026.