Last updated: October 2026
These guidelines govern everything AI Briefing News publishes. They are written to be specific enough to enforce, because vague standards are unenforceable.
Who we serve
We write for people who work with AI — engineers, founders, executives, product and policy professionals — and for readers who need to understand it without working in it. We assume fluency in AI fundamentals in our news and analysis, and we assume none in our explainers.
What we publish
We publish four distinct formats, and we do not blur them.
- **News** reports what happened, factually, without interpretation.
- **Analysis** explains what a development means, grounded in cited evidence, and always includes the strongest case against its own reading.
- **Explainers** teach a concept from first principles and carry a visible last-reviewed date.
- **Opinion** argues a stance, is labelled unmistakably as opinion, and runs only under a real named author.
Sourcing
We grade every source before we rely on it. Primary documents — regulatory filings, court rulings, published standards, technical papers — and wire services carry the most weight. Reputable trade press and credible analyst notes come next. Vendor announcements, company blogs, aggregators and unattributed posts carry the least, and never support a consequential factual claim on their own.
**We treat company announcements as claims, not facts.** A vendor launch post is a starting point for reporting, never a substitute for it. If the only thing we can do with an announcement is restate it, we do not publish it.
Where a claim is single-sourced and we could not independently corroborate it, we attribute it explicitly in the story rather than presenting it as established fact.
Verification
Every factual claim we publish traces to a source we actually consulted. We do not fabricate quotes, figures, dates, names or affiliations under any circumstances.
Where a detail is undisclosed or ambiguous, we write that it is undisclosed. We would rather appear uncertain than be wrong.
We also treat what a source leaves unsaid as reportable. Undisclosed pricing, unspecified availability, and benchmark claims published without methodology are routinely the most useful lines in a story.
We distinguish fact from inference in our language. "This suggests" and "this proves" are not interchangeable, and in analysis every interpretive claim is marked as interpretation.
Copyright and attribution
We paraphrase into our own words and structure rather than reusing a source's phrasing. Direct quotes are kept short, placed in quotation marks, attributed to a named speaker and linked to the source. We do not reproduce another publication's paragraphs, lists, tables or charts.
We do not use images, screenshots, diagrams, figures or video stills taken from other sources. We commission or licence our own visuals, or we publish none.
Where a statistic originated with a third party that our source was citing, we credit the originator rather than the article we read it in.
How we use AI
We cover artificial intelligence and we use it in our own newsroom. It would be strange to hide that.
AI assists with research, source gathering and drafting. It does not make editorial decisions. A human editor determines what we cover, verifies the facts, and reviews every piece before publication — and holds an AI-assisted draft to exactly the standard we would hold any reporter's.
We do not publish unreviewed machine output. We do not generate quotes or sources. We do not present AI-generated imagery as real. When we get something wrong, a named person is accountable for it and we correct it in public.
Bylines and accountability
News, analysis and explainers carry either a named author or the AI Briefing News Newsroom byline. Opinion always carries a real, named author — never a collective byline — and the position argued is one that author actually holds.
Fairness
We attack positions and arguments, never people. We do not publish claims about a named individual's motives, competence or character that we cannot source. When we quote someone in order to disagree with them, we quote them fairly and in context.
Where a story is materially critical of an organisation, we seek comment before publication and record whether it was given.
Independence
Our commercial activity has no influence on editorial decisions. Sponsored content is labelled as sponsored. Directory listings, event listings and rankings are editorially determined and are not for sale.
Corrections
When we are wrong, we append a dated correction to the piece. We do not quietly edit the record. The full process is set out in our Corrections Policy.