These are the rules we hold ourselves to. If you see us break one, tell us at contactus@gcwhistleblower.com and we will answer in public.
-
We start from the record
Reports are built on public documents: agendas, packets, minutes, budgets, contracts, bid results, court filings, election results and official notices. Each report links to the documents it relies on. When we rely on another news outlet or an advocacy group, we say so and name it.
-
We ask before we publish
Before we publish a report that raises a question about a named person, office, business or organization, we ask them for comment and give them a reasonable time to answer. The report then says one of three things: what they said, that they declined, or that they did not reply by our deadline.
Read the rest of rule 2
Not every report raises such a question. Service reports that only summarize a public agenda, packet, minutes, bid result or election record, such as what a board will vote on or what a project costs, do not question anyone’s conduct, and we publish them without a request for comment. Many of them end with a note that the offices named have not been asked and are welcome to respond. That note is an open invitation, not a request: anyone named can use it, and a reply is published in full.
If a report that does raise a question has to go out before we can ask, for example because of a same-day deadline, it says plainly that comment was not requested. We then send the request after publication, list it on the Requests log, and add any reply to the report with a dated note.
We also publish the asking. Every request for comment we send to a public official or agency, and every reply, is listed with its full text on our Requests for Comment Log, with the date sent and what came back. Each request tells the recipient this up front. We remove only personal contact details such as home addresses and personal phone numbers. The log lists every request actually sent: if a report names an office and the log has no entry for it, we have not sent one.
This rule took effect Oct. 1, 2026. Reports published before then were written without it, and some do not say whether comment was requested.
-
Anyone named can answer, at any time
A response sent to contactus@gcwhistleblower.com is published in full, with the report it answers. There is no deadline for a reply.
-
We label what is not settled
Confirmed facts are stated as facts. Claims, disputes and open questions are labeled as such. Most reports end with a short list of what we could not establish and which records would answer it. A report still in progress is marked “Developing” and carries an update history.
-
We do not guess at motives
We report what a person did or said and what a record shows. We do not state why someone acted unless they told us or a record says so.
-
We correct in public
Published reports are not quietly rewritten. An error is corrected with a dated note, and every correction is listed permanently on the Corrections & Updates page.
Read the rest of rule 6
When possible, each entry says what was published, what is wrong or changed, when, and why. The original wording stays as published. A spelling, grammar or broken-link repair that does not change the meaning may be made without a notice; any change to a fact, a figure, a name or an attribution is always disclosed. To ask for a review, write to contactus@gcwhistleblower.com with the report, the sentence in question and the record that shows the problem. We answer in public.
When something changes after a report runs, for example a vote that has since taken place, the report gets a dated update note and an entry on the same page. A script flags reports that still describe something as upcoming after its date has passed, so the notes can be added.
-
News and opinion stay apart
Our reports are news. If we ever publish opinion, it will be labeled as opinion.
-
Scanner traffic and social media are tips, not facts
Dispatch radio and online posts are often wrong at first. We report them only with clear attribution, and we look for the official record.
-
Private people and children
We do not publish home addresses, phone numbers or medical details of private individuals. We do not name children in connection with wrongdoing. Public officials are covered for what they do in office, not for their private lives.
-
Photos
We use only images we are allowed to use: public-domain and openly licensed photographs, public records, government aerial imagery, stock photos labeled as stock, and our own charts. Every image is credited, and a stock or file photo is labeled so it is not mistaken for the event.
-
Anonymity and sources
The people behind this publication are anonymous. We do not reveal who gives us information unless they agree. We do not publish an anonymous accusation on its own; a tip has to be backed by a record or a named account before it is reported as fact.
The reasons, in short: to protect the people who talk to us, to keep attention on the records and not on whoever wrote about them, to ask the same question of anyone without fear of payback in a small county, and because the law allows it. Anonymity has costs, and the About page names each one and what we do about it. Because you cannot check who is behind this site, we make the work checkable instead (rules 1, 6 and 16).
-
Public representatives
Volunteers who contact offices for us do so under their own names and always say they represent this publication. Their role is described here.
This is how we ask questions and file records requests without a name of our own: a volunteer signs the request or asks the question in person or in writing, says whom they represent, and forwards every answer to us unchanged. A volunteer is named in public only with their own permission. Each request is listed on the Requests log with its full text.
-
Community posts
The achievements feed is for good news. Posts are checked automatically and appear right away, and the people behind this publication remove any post that breaks the submission policy. Reader photos and questions are published only with the sender’s permission.
-
Privacy
We collect as little as we can. Details are on the privacy page.
-
Audio versions
Reports can be read aloud by an artificial voice, labeled “AI voice” wherever it appears (see also rule 18). The audio is made from the published text and is regenerated whenever that text changes, so it never gets ahead of or behind the report. Our two reading voices are named in remembrance of Koda and Queenie, the two dogs from our report. Read that report, or see why we named them.
-
How a report is checked
Every figure, date, dollar amount, quotation and name in a report is meant to come from a record the report links. A checking script reads each linked page and document and looks for them. It flags a figure that is not in any linked source (a figure we calculated, such as a difference or a percent, passes only when the numbers it comes from are in the linked records); a headline or summary figure that disagrees with the report; a quotation or a name that is not in a linked record; a date that has passed after wording such as “will vote”; a weekday that does not match its date; loaded words; and statements of motive that no record supports.
Read the rest of rule 16
Some sources cannot be read by a script: scanned documents, video, and pages that load their content with scripts. The script marks those “needs a human check” and never counts them as verified.
The script runs on every new report, and it was run on every earlier report on Oct. 3, 2026. A published report is not quietly fixed when it finds something: the published text does not change, and a clear error is corrected with a dated note on the Corrections & Updates page (rule 6).
-
Sensitive reports
Some reports name a person or a public body in a critical context: a denial, a missing record, a violation, a complaint, a question about conduct. Software flags these as sensitive. A sensitive report is held out of automatic posting and the weekly newspaper until the person who runs this site has approved it in the site’s records. The request for comment that rule 2 requires must have been sent first and be listed on the Requests log.
This check began Oct. 3, 2026. Reports published before then were not held this way.
-
How we use AI
We use artificial intelligence as a tool. This is what it does here, and what it does not.
What AI helps with. Research: finding and reading public records such as agendas, packets, minutes, bid results and statutes, and pulling out the figures, dates and quotations in them. Checking: comparing the figures, dates and quotations in a report with the records the report links (rule 16). Audio: our two reading voices, Queenie and Koda, are AI voices, labeled “AI voice” wherever they appear, and they narrate our short videos. Images and video: the share cards and Reels are generated by software from a report’s published text and figures.
What we do not do. We do not publish AI-generated photographs, and we do not present any generated image as a photograph; photos are real, credited and labeled (rule 10). We do not invent quotations, sources or facts. Every figure in a report is meant to tie to a public record the report links, and an AI tool is never cited as a source.
What stays with people. A person sends every email and posts every reply to a reader or an official. Software never emails or messages anyone.
How you can check. Each report lists the records it links, and the “AI assistance” line under its headline says which of these uses applied to that report. Open the sources and compare. If a figure does not match its record, tell us; we correct it in public with a dated note (rule 6).
This describes how the site works as of Oct. 3, 2026. If it changes, we say so on the Corrections page.
-
Who runs this site and who pays for it
This site is not part of any government, party, campaign or news organization. The people behind it are anonymous (rule 11) and unpaid. We do not accept money from readers: no donations, memberships or subscriptions, and no paywall. We do not accept payment for coverage from a business, a candidate, a party or a public body, and we do not sell or publish legal notices.
The only income we plan is advertising that an ad network places automatically. It is labeled as advertising and kept apart from the reporting, and advertisers have no say in what we cover. As of Oct. 2, 2026 the site had earned no money and was not a registered business. The Support page and the About page carry the current status and say first if anything changes.