Editorial transparency
How AI fits into our editorial process
We use AI in research, drafting and lint checks - and we're open about it. This page covers exactly where, what editorial oversight looks like, and the lines that don't move regardless of how good the tooling gets.
0
Fabricated reviews or quotes - ever
100 %
% of pages editorially reviewed
1:1
AI assists, an editor decides
Where we use AI
Specifically, large language models (currently from Anthropic and OpenAI) - used as production tools by a human editor.
Research synthesis
Aggregating manufacturer specs, official guidance and authoritative third-party sources into a structured brief before a piece is written. The brief is read by an editor; it isn't the piece.
Drafting
Producing first-draft prose from the brief - then edited line-by-line, fact-checked against the brief's sources, and rewritten where it doesn't sound like us.
Internal linking and SEO checks
Identifying related pages on the site for cross-linking, and running on-page lint checks for metadata length, heading hierarchy, image alt text, and slug style. The lint suite catches problems; the editor decides what to fix.
Content audits
Spotting outdated claims, broken links, and content that needs refreshing across the site at a scale a single editor can't keep up with manually.
The rules that don't move
These apply equally to AI-drafted and fully human-written content. Tooling changes; the lines don't.
We will
-
Disclose AI assistance when material
If a draft was AI-assisted in a way that shaped the page's content, that's disclosed in our editorial process - this page is the disclosure.
-
Fact-check every claim against a source
Prices, regulations, manufacturer specs, accommodation policies - checked against the operator or vendor at publication time, not against the model's training data.
-
Edit until it sounds like us
AI drafts pick up patterns from training data. We rewrite anything that reads like a remix of competitor pages or generic-internet-blog prose.
-
Take editorial responsibility
The byline is on us. "The AI wrote that bit" is not an excuse we will ever offer for an inaccuracy or a tonal misstep.
We will not
-
Fake first-person experience
We never write "we stayed at this hotel", "we tested this harness", or "we visited this beach last summer" unless that is genuinely true. Our content is editorial - research-led, not experience-faked.
-
Fabricate reviews or testimonials
We never invent customer quotes, fake user reviews, or fabricated testimonials from people who do not exist.
-
Invent statistics or studies
We don't make up percentages, survey results, or study figures. Where we don't have hard data, we say so - or use hedging language ("typically", "in many cases") rather than a precise-sounding number.
-
Quote users from forums
Real people on Reddit, forums and customer-review pages did not post their comments for republication on a commercial site. We read the patterns of community discussion as research input - we do not quote, attribute, or link to individual users.
-
Fabricate credentials
We don't invent expert authors, claim affiliations the team doesn't have, or pretend the site is staffed by people who don't exist.
Editorial oversight
Every piece published on Four Legged Guests is reviewed before it goes live. The review covers four things.
Editorial oversight, in four passes
Every page - AI-drafted or not - runs through these four checks before it publishes.
Factual accuracy
Claims that look out of place - wrong product specs, outdated regulations, accommodation policies that have changed - get checked against the source before publication.
Honest framing
Anything that reads as first-person experience or implied expert authority is removed or rewritten. "We found that" becomes "reviewers report"; a confident claim becomes a hedged one when the evidence is thin.
Originality
AI drafts can pick up patterns from training data. We rewrite anything that looks like a remix of a single competitor source, and we add the angles and warnings other pages don't include.
Affiliate compliance
Affiliate links are disclosed, marked with the right HTML attributes, and never hidden behind language that suggests they're not commercial. The disclosure rule is the same regardless of who wrote the page.
Why we tell you this
The honest reason is trust. Readers who know how content is made make better decisions about what to act on. Disclosure costs us nothing and is the standard we want to set for the niche.
The shorter version: AI is part of how the site is made, editorial oversight is non-negotiable, and the rules above are not aspirational - they're the lines we won't cross regardless of how convenient it would be to.
AI assists. An editor decides. If a page is wrong, the responsibility is ours — not the model's.
Got a question about how we work?
If you want to know how a specific piece was researched, or you've spotted content that crosses any of the lines above, please get in touch. Our full editorial policy covers our broader commitments.