Our position
Plenty of sites use AI and say nothing about it. Others promise they never touch it, which is increasingly hard to believe. We would rather tell you the truth: AI tools are part of how this site is produced, in defined roles, under human control.
The principle we work to is simple. AI can help a person work faster. It cannot be the person who is accountable. Every article has a human name on it, and that human is answerable for every claim in it — whether or not a machine helped produce the first draft.
Where we use AI
- Research assistance. Summarising long documents, locating relevant studies, and surfacing questions we had not thought to ask.
- Drafting. Producing a first draft or restructuring an existing one, which a human then verifies, rewrites and takes responsibility for.
- Editing support. Tightening prose, catching inconsistencies, flagging passages that overstate certainty.
- Routine production work. Alt text suggestions, meta descriptions, formatting and other mechanical tasks.
Where we never use it
- Deciding a verdict. Whether a product is recommended is a human judgement informed by using the thing.
- Standing in for testing. A model cannot tell you whether a clip loosens after six weeks. Only using it can.
- Generating sources. Every reference is a real document a human has opened. Language models are known to fabricate plausible-looking citations, so we never accept a source we have not checked ourselves.
- Passing as a person. No AI-generated byline, no invented author, no fake reviewer. Every name on this site is a real person — see our team page.
- Final health sign-off. Veterinary review is done by a qualified human, as set out in our medical review policy.
What a human always does before publication
- Reads the article in full.
- Verifies every factual claim against a primary source, independently of whatever the draft asserted.
- Opens every cited reference and confirms it exists and says what the article claims it says.
- Removes anything that cannot be substantiated.
- Puts their name to it.
If those steps have not happened, the article does not publish. There is no route by which unreviewed machine output reaches this site.
Why this matters for accuracy
Language models produce fluent, confident text regardless of whether the underlying claim is true. That fluency is exactly what makes unchecked AI content dangerous in a subject where being wrong has consequences — a bad feeding recommendation or a missed symptom affects an animal that cannot tell you something is wrong.
So we treat AI output as a draft to be verified, never as evidence. The verification step is the work. The drafting is the easy part.
How we label AI involvement
Where AI has materially contributed to an article's drafting, we are happy to say so — being asked is not something we consider an accusation. If you want to know how a specific article was produced, email [email protected] and we will tell you.
If you spot something that looks machine-generated and wrong, that is a failure of our process, and we want to know. Report it here — it will be treated as a correction, not a complaint.