for agents/llms.txtv0.14.14

Home / Articles / Becoming an official publication: the SGit Newsroom now has a publisher, an editor, standards, a corrections log, a withdrawal process and a complaints route, and is applying to a regulator

Becoming an official publication: the SGit Newsroom now has a publisher, an editor, standards, a corrections log, a withdrawal process and a complaints route, and is applying to a regulator

By · 2026-10-30 · article v1.2.0, 3 versions · newsroom v0.14.11 · newsroompublishingjournalismstandardsregulationcorrectionsai-disclosurearticle

Abstract: The SGit Newsroom has been publishing for weeks, and it names companies, reports incidents and argues positions, written by agents from my voice memos. Today a story showed me that this needs more than good habits. We wrote about a contractor's agent on a Pentagon network from secondary reports of an article we could not read, copied a quote that one outlet had given to the wrong party, corrected it, and then I withdrew the article, because it repeated allegations about named organisations made by unnamed sources, which this newsroom should not publish at all. So the newsroom is becoming an official publication. In the UK there is no licence to be a journalist; what makes a publication accountable is that it says who publishes it and who answers for it, publishes its standards, corrects in the open, withdraws with a notice, takes complaints, and submits to an independent regulator. As of this release the newsroom has an imprint, naming The Cyber Boardroom Limited as publisher and me as editor, and saying plainly that agents write the articles; editorial standards that rule out making or repeating allegations, so what we publish is analysis and opinion; a corrections log; a formal withdrawal process with its first notice; and a complaints route with a 21-day answer. Next we apply to IMPRESS, the independent press regulator, and register for an ISSN with the British Library, and later we look at the Journalism Trust Initiative and a press card for the editor. This article says what each step is, what it costs, what we have done and what is still to do.

Four steps to an official publication, and where the SGit Newsroom stands on each, 30 October 2026.

In short

Why now

This newsroom has been publishing for weeks, written by agents from my voice memos and briefs, under rules that are published and checked by the build. It names companies, reports incidents and takes positions. Today a story showed me that good habits are not the same as being accountable.

We wrote about WIRED's report of a contractor's AI agent on a Pentagon network. WIRED's page refused our tools, so we read it through what six other outlets said about it, and said so. When I later shared the article itself, it showed that one outlet had joined the end of the Department of Defense's statement to the start of Accenture's, and we had copied it. We corrected it within the hour. Then I withdrew the article, because it repeated allegations about named organisations made by unnamed sources, and that is not something this newsroom should publish at all. That process is what an official publication has, and what we did not have written down.

What makes a publication official

In the UK there is no licence to practise journalism, and no register of journalists. What makes a publication accountable is a set of things it does in public:

On top of these sit identifiers and credentials: an ISSN from the British Library for a serial publication, certification such as the Journalism Trust Initiative, and press cards, which are issued to people, not publications.

What we did today

The newsroom now has five pages that it did not have this morning, linked from every page's footer under "The publication":

Most of these write down what the newsroom already did: every source dated, quotes verbatim, every version public, reviews on the record. Two things are new rules. First, opinion and not allegations: we do not accuse any person or organisation of wrongdoing, and we do not repeat other people's allegations either, because in UK law repeating an allegation is publishing it. What we publish is analysis and my opinion, said in the first person, with the facts it rests on, so a reader can see the basis and disagree; and when we criticise a named person or organisation, we let them know and publish their response. Second, withdrawal: only the editor withdraws an article, and every withdrawal gets a notice.

Agents write here, and the imprint says so

The question every reader will ask of this newsroom is who writes it. The imprint answers it in one paragraph: the articles are written by AI agents working as the newsroom's writers, from my memos, briefs and the material I share, under the published standards and each agent's published behaviour policy, and I direct and answer for every article. Nothing is published that I would not say.

I think that is the right way round. An agent can research, check and write faster than I can, and it does not tire of looking up a source. But it cannot answer for what it publishes. A person has to, and the imprint names who.

The first withdrawal

The withdrawal process is new, so its first notice is about today. The notice gives the article's title, when it was published, corrected and withdrawn, with the versions, and the reason. The article's number stays in the newsroom's register, marked withdrawn, so a reading list that named it says so. The source stays in the public repository's history: a withdrawal removes an article from the site, not from the record. If we return to the story, it will be a new article, after the parties have been asked.

The withdrawal process: the editor decides, the article leaves the site, its number is kept and a notice is published.

What comes next

Is SGit Newsroom the right name?

I think so. It says where the newsroom comes from: it was built on sgit, the tools and the vaults it writes about, and it is open about that. The risk is that a reader takes it for a product blog. The imprint and the standards are the answer to that: the publisher, the editor, the interests and the rules are on the page, and any article about something we build says so.

What this does not claim

Where this comes from

Peer review Not reviewed yet. Review it

Threads

News & evidenceSite & engineering This article as a graph →

Builds on

All articles · All graphs

Article No. 334 · start a reading list with it

Want the next issue by email. One issue a week or so: what was published, what it adds up to, and what is worth your time. Subscribe to the SGit Newsroom →

← All articles

Reviews

Reviews on the record, by agents, the editor and the desk, against the criteria. Each is kept for good and answered finding by finding. All reviews →

No reviews yet.