Public interest immunity is the mechanism by which a government declines to give the Senate documents it has ordered. Individually, each claim is a judgement call. Collectively, the series is one of the few hard measures of executive openness that does not depend on anyone's opinion.

The Senate does this job properly. Its consolidated list for the parliament records, for each order: the subject, who moved it, the date agreed, the date and nature of any response, the grounds of any public interest immunity claim set out as a list, and a compliance status drawn from a defined set -- complied with, partially complied with, no documents exist, not yet complied with -- each of which the document defines. Grounds appear in the form a reader can work with, for example "disclosure of privileged legal advice" or "prejudice to the investigation of a possible breach of the law".

The gap left is narrow and worth naming anyway. It is one PDF, so establishing how claims are trending, which portfolios account for them, or how compliance compares across parliaments means extracting several hundred entries by hand. That work currently gets done by advocacy organisations, which then lets the resulting picture be discounted as advocacy -- a poor outcome for a Senate whose own record is good.

Read the Senate's consolidated orders for the production of documents list for the parliament, which is kept current and carries everything described above. For a series, transcribe it.

The Centre for Public Integrity periodically compiles and analyses the same material, which is where most published comparisons across parliaments come from.

Marked verified on 2026-09-03, with one figure corrected downward in the process. This entry previously said the consolidated list "contains 284 occurrences of the phrase public interest immunity". That counted every mention, including the guide at the front explaining what the term means. The number that answers the question is how many orders carry an itemised claim, and that is 64.

The error inflated the finding in the register's own favour, which is the direction it should be most suspicious of. Counting a phrase is not the same as counting the thing.

Checked on 2026-09-01, and the check overturned this entry.

The earlier version claimed the collated table did not exist and that assembling one was research work nobody had done. That was wrong. The Senate publishes exactly such a table. The consolidated list for the 47th Parliament was downloaded and searched: 64 orders carry a "Public interest immunity claims raised" heading with the grounds itemised beneath it, and a guide at the front defining each compliance status.

Recording this correctly matters more than the stronger claim did. The register exists partly to identify where Australian transparency works so the failures have a benchmark, and an entry asserting a gap that a parliamentary office had already closed would have been a false gap of the worst kind -- one that disparages a body doing the thing the register wants done.

What is left is a genuine format problem and nothing more, so the entry claims nothing more. The ask has shrunk from "build a register" to "publish the register you already maintain as CSV alongside the PDF". Every blocker is a publishing choice, so this remains fixable without new collection -- now for a much smaller and more obviously reasonable request.

Still worth checking: whether the Senate Table Office publishes any structured extract, and whether the compliance statuses are applied consistently enough to count on.

Related: Which agencies refuse the most FOI requests, and on what grounds?, Which lobbyists and organisations met which ministers, and about what?, Commonwealth FOI.