Artificial intelligence is now one of our biggest beats, and it's also reshaping how information gets made across the whole industry. Our readers deserve to know exactly where we stand. This is our policy on the use of AI in Circuit Bulletin's journalism, and it applies to all of our writers, editors, and contributors.
In short: our reporting is the work of people, from the first draft to the final edit. We don't hand our writing, our images, or our reporting decisions to a machine, and we won't attribute anything to a source that a model has conjured up. We're not a large outlet, but we've watched a lot of the industry lean into AI-generated content and decided it isn't for us. Our approach to human-authored content is not going to change.
Our journalism is human-authored
Every word of our editorial text is written by a human being. We do not use AI to write our reporting, analysis, or commentary, not in whole and not in part. When we're reporting on AI itself and it's relevant to show what a model produced, we may reproduce that output as an example, clearly labelled as AI-generated and set apart from our own words. That's the subject of the story, not our writing.
No synthetic sourcing
When we attribute a quote, a position, or a view to a named person or organisation, that material comes from real engagement: an interview, a transcript, a published statement, or a document a reporter has actually read. We never let a model produce, pull out, or condense anything that then gets pinned on a named source, be it a quotation, a paraphrase, or a summary of what someone thinks. We don't publish claims that rest solely on an AI summary, and no one may describe a document as "reviewed" unless they've examined it directly. This is a bright line, and it connects directly to the ban on fabrication in our Ethics Policy.
Images, audio, and video
Our visuals are made by people. Illustrations, graphics, and video come from our own team or from photographers and wire providers, guided by human creative judgement. We do not publish AI-generated images, audio, or video as authentic documentation of real events, and we do not alter documentary media in ways that change what it shows. Ordinary production work such as cropping, colour correction, and contrast adjustment is fine. On the rare occasion synthetic media appears in the course of reporting on AI, it's clearly identified as such, with the disclosure placed as close to the material as possible.
Where we permit AI tools
We're not pretending these tools don't exist, and we won't ban a spell-checker on principle. We may use AI-powered tools for narrow, supporting tasks: checking spelling and grammar, suggesting stylistic tweaks, and offering structural feedback on a draft. The crucial limit is that a tool can only ever suggest. A human reads every suggestion and decides whether to act on it, and a human actions every change. The tool never has the final say, and it never has any say over the substance of what we report.
Human review
Every article we publish is reviewed by a human editor before it goes live. Responsibility for accuracy and integrity rests with the people who make the work, and it can't be handed off to a colleague, an editor, or a piece of software. Where a story needs specialist checking, we use human subject-matter experts, as described in our Editorial Guidelines.
Why we take this position
We're sceptical of AI in journalism for reasons of substance, not nostalgia. Large language models are built to produce plausible text, not true text, and they routinely hallucinate: inventing facts, sources, quotes, and citations that look convincing and are simply false. A model has no understanding of what happened, no way to stand behind a claim, and no accountability when it's wrong. Journalism is a chain of human judgement and human responsibility, and handing any link of it to a system that cannot be held responsible breaks the very thing that makes reporting trustworthy.
There are wider harms too. AI-generated content is often trained on the uncredited work of writers, artists, and photographers, it can dilute the information ecosystem with confident nonsense, and it erodes the trust between a publication and its readers. We'd rather be slower, more careful, and unmistakably human. When you read Circuit Bulletin, a person researched it, a person wrote it, and a person is answerable for it.
