AI Usage Policy

by Andrea Hartley
/ ✕ What we use AI for. What we never use it for. Disclosure rules.

AI Usage Policy

Last updated: May 2026 · Author: Eleanor B., Lead Editor

Gardening content on the open web in 2026 is increasingly produced by language models — lightly edited, dressed up with stock photography, and presented as written by experienced gardeners. We do not do this. This page is the explicit, public commitment to what we use artificial intelligence for, what we never use it for, and how we disclose any AI involvement when it occurs.

The policy applies to everything published on Explore Your Garden: how-to guides, plant profiles, tool and product reviews, seasonal calendars, the glossary, the beginner’s guide, and the static informational pages. It also applies to images, video, and any other media we produce.

1. What AI is never used for

  • Generating editorial drafts. No plant guide, how-to article, or review is drafted by a language model and then edited into shape. The first draft of every long-form piece is written by a named human team member who has done the underlying gardening work.
  • Generating opinions or recommendations. Our verdicts on plants, tools, and practices are formed by humans who have actually grown, used, or worked through the subject. We do not ask language models to “write a review of X tool” and then dress it up in our voice.
  • Generating fake authorship. Every byline corresponds to a real person on our Meet the Team page. We do not invent fictional contributors to disguise AI-generated content.
  • Generating fake garden photographs. We do not use AI image tools to generate photographs of plants, gardens, or tools and present them as documentary evidence of our gardens or our test results. Real plants get real photographs.
  • Generating fake plant identifications. Plant ID and foraging articles in particular are never AI-generated, because the consequences of bad plant ID can be serious. See Plant Safety & Hazards.
  • Inflating word counts. Articles are as long as they need to be to do their job and no longer. We will not pad them with AI-generated filler to hit a length target.

2. What AI may be used for

Some narrow operational uses are permitted, transparently:

  • Translation assistance for botanical Latin and non-English sources. When a non-English horticultural reference contains a useful term, we may use AI translation as one input alongside dictionaries and human judgment. Final word choice is always made by the writer or editor.
  • Spelling and grammar checking. Standard spell-check and grammar tools are not controversial.
  • Organizing personal garden notes. A team member may use AI to help structure their own field notes into outline form. The outline still has to survive editorial review, and the writing from the outline is still done by the human.
  • Image enhancement. Standard photo-editing tools, including AI-assisted enhancements (sharpening, noise reduction, color correction), are used to make our garden photographs presentable. We do not generate new visual content this way and do not alter photographs in ways that misrepresent what was in the garden.
  • Search and retrieval. AI-powered search tools that help us find relevant past articles, sources, or facts are useful and uncontroversial; what they surface is verified by a human before being published.
  • Accessibility. We may use AI-generated alt-text drafts as a starting point, with every alt-text reviewed and rewritten by a human before publication.

3. Disclosure rules

  • Routine uses (spell-check, photo enhancement, alt-text drafting, search) require no per-article disclosure. They are covered by this policy page.
  • Non-routine uses on a specific article (translation help on a substantial passage, AI-organized note structures) are disclosed in a note at the end of the article in plain language: “AI tools used in producing this piece: [list].”
  • Future expansion of permitted uses is reflected in updates to this page, with a changelog at the bottom showing what changed and when.

4. Why we draw the line where we draw it

Gardening writing has a specific failure mode that AI text exposes brutally. A language model trained on a thousand articles about growing tomatoes can write a perfectly fluent piece on tomato cultivation. But the value of a tomato guide is not the fluent assembly of common knowledge. The value is the writer who can say, “I have grown this variety three years running in heavy clay; the standard advice about pruning suckers is wrong for indeterminate varieties in cool maritime conditions, and here is what works instead.”

Language models cannot say this, because they have not done it. They can produce text that sounds like someone who has, but the underlying experience — the watch through the season, the specific failure observed, the specific correction discovered — is not there. In gardening, this gap is the difference between writing that helps and writing that wastes the reader’s time.

We are also acutely aware that AI-generated gardening content has caused real harm in the foraging space, where confidently wrong identifications have appeared in machine-generated guides and have, in documented cases, led readers to consume toxic plants. We will not contribute to that risk.

5. Reader rights and recourse

If you suspect that an article on Explore Your Garden was generated by AI in violation of this policy, please email info [at] exploreyourgarden [punto] site with the subject line AI policy concern and the URL of the article. We will investigate, and we will publish a public response if any violation is confirmed.

If you would like to use one of our articles as part of training data for an AI model, please see our Copyright Notice and Terms of Service. The short version: we do not authorize use of our content for training of generative models without explicit written permission.

6. Future-proofing this policy

The technology is moving quickly, and a policy written in mid-2026 may need updating as new tools emerge and editorial practices evolve. We commit to:

  • Reviewing this policy at least once every six months.
  • Maintaining a visible changelog when the policy is updated.
  • Treating any expansion of permitted uses as a serious decision requiring team consensus, not a quiet edit.

For our position on the related questions of editorial standards, conflicts of interest, and reader trust more broadly, see Editorial Standards. For our process, see How We Work.

Related pages: Editorial Standards · How We Work · How We Test · Copyright Notice · Sources & Citations