AI Disclosure on Articles: What to Say Without the Lawyer Fog

By Austin Little

If a machine helped draft the page you are reading, say so in plain English. Not a wall of legalese. Not a wink. A short label that tells people what happened, what you checked, and what you did not. This is the Apiary playbook for AI disclosure on articles — what to put, where to put it, and what to skip.

AI disclosure. This page was drafted with AI assistance and edited for Apiary. We don't invent quotes, stats, people, or events. If something looks off, tell Austin — that's the point of a living hive.

Why disclosure exists (and why fog ruins it)

People already assume a lot of web writing is machine-touched. That does not mean you get a free pass to hide it. Disclosure is not a confession booth for shame. It is a trust tool. It says: here is how this page was made, so you can judge it the same way you judge any other source.

Fog shows up when writers try to sound more careful than they are. Long paragraphs about "may or may not have utilized generative systems in certain portions of the drafting pipeline" teach the reader nothing. Fake precision shows up the other way: "47% of sentences were AI-generated" when nobody counted. Both are worse than a short honest line.

Apiary's bias is simple:

Say AI helped if it did.
Say a human directed and edited if that is true.
Do not invent quotes, stats, people, or events to fill gaps.
Invite correction. A living hive beats a perfect-looking lie.

If you want the free tool side of writing with machines, see How to Use AI Without Paying. If you want the habit of tending knowledge with oversight, see What Is an AI Beekeeper?.

What readers actually want to know

When someone skims a disclosure, they are usually asking five questions:

Did a machine write part of this? Yes / no / mixed.
Did a human read it before publish? Yes / no / unknown.
Were facts checked against primary sources? Fully / partially / not yet.
Is this advice I should stake money or health on? Usually no — treat as research notes unless a licensed human signed it.
Who do I yell at if something is wrong? A named person or inbox.

If your disclosure answers those five, you are ahead of most of the internet. If it answers none of them and only says "AI may have been used," you wasted the reader's time.

The Apiary short block (copy and adapt)

This is the block we use on craft pages. Steal the shape. Change the names.

> **AI disclosure.** This page was drafted with AI assistance and edited for Apiary.
> We don't invent quotes, stats, people,
> or events. If something looks off, tell Austin — that's the point of a living hive.

Why this shape works:

Bold label so skimmers catch it.
One sentence on process (drafted with AI, edited by humans / for the site).
One sentence on rules (no fake facts).
One sentence on accountability (who to contact).

You can put it under the byline, at the top of the article body, or both. Top is better for trust.

What to put — a checklist you can actually use

Must say

That AI assisted drafting (or that it did not, if you want the rare "human-only" badge).
That a human edited / directed (if true).
That important claims should be verified (especially medicine, law, money, safety).
How to report an error (email, form, or named person).

Should say when relevant

Date of fact-check or "checked as of [month year]."
Which kinds of facts are flagged (model names, prices, legal thresholds, seasonal bee advice).
Whether quotes are from real people / docs vs paraphrased.
Whether images are AI-generated, stock, or photos you took.

Usually skip

Token counts, model version strings, and "temperature=0.7" theater unless your audience is researchers who asked.
Fake percentages of AI vs human sentences.
Threats ("by reading you agree…") that belong in a real Terms page, not a craft article.
Naming every tool in the chain if the chain changes weekly — "AI assistance" plus a human editor is enough for most readers.

Lawyer fog vs plain English (side by side)

Fog: "Portions of this content may have been generated, refined, or otherwise influenced by automated systems including but not limited to large language models, and readers acknowledge that such systems can produce inaccurate outputs."

Plain: "AI helped draft this. A human edited it. Do not treat this as legal or medical advice."

Fog: "Any resemblance to real persons or events is coincidental unless otherwise indicated in accordance with applicable policies."

Plain: "We do not invent people, quotes, or events. If a name appears, it should be real and checkable — or it is a clear fictional example labeled as such."

Fog: "The publisher disclaims all warranties regarding completeness…"

Plain: "We can be wrong. Tell us. We will fix it."

You can still have a real legal page for the site. The article disclosure is for humans reading a how-to, not for a courtroom exhibit.

Where platforms and laws fit (without pretending we are your counsel)

Rules move. Platforms invent labels. Some regions talk about "AI-generated content" marks. Search engines publish quality guidance that rewards helpful, people-first pages and punishes scaled spam. None of that replaces your own honesty policy.

Practical stance for a small publisher like Apiary:

Follow the platform's required labels when you post there (social, marketplace, app store — whatever they demand that week).
Keep your own site disclosure clearer than the minimum.
Do not claim compliance certifications you did not earn.
Do not invent a statute citation from memory.

This article is not legal advice. It is a craft standard: say what you did.

Disclosure for different article types

How-to craft (this batch)

Use the short block.

Bee / habitat / science-adjacent

Add an extra line: treatments, doses, and legal hive rules must be verified with a mentor, club, or extension source — AI can invent confident nonsense about mites and meds.

Local service / garage / money

Add: not a substitute for a licensed contractor, insurer, or attorney. Quotes and prices change. Verify licenses yourself.

Say it is opinion. If AI helped polish, still disclose. Readers forgive opinion. They do not forgive fake "studies."

News-ish roundups

If you summarize other outlets, say you summarized and link them. Do not let the model invent a quote from a press conference that did not happen.

How to mark uncertain facts without killing the read

In drafts we use HTML comments:

On the live page, convert the important ones into visible hedges:

"As of September 2026, check the provider's free-tier page — limits change."
"Model names churn; pull a current small model from the project's own docs."
"Confirm the license threshold on CSLB's site before you rely on a dollar figure."

Visible hedges beat silent guessing. Silent guessing is how fake facts sneak in.

What disclosure is not

Not a free pass for spam. Labeling AI sludge does not make sludge good.
Not a substitute for editing. "AI wrote it, deal with it" is not a brand.
Not a gotcha for readers. You are informing them, not tricking them into clicking Accept.
Not a claim that humans are obsolete. The point of Apiary-style work is human direction plus machine draft speed.
Not a watermark detector arms race. Public "AI detectors" are unreliable for soft prose. Do not hang your ethics on a score from a random scanner.

If someone demands a detector percentage before they trust you, show your sources and your process instead. Process beats magic meters.

A one-page disclosure policy for your own site

Write this once. Stick it in /disclosure or your about page. Point article blocks at it.

Default: AI may assist drafting on craft and knowledge pages.
Human gate: Nothing important publishes without a human pass (or a marked draft status).
No fake facts: No invented citations, interviews, or statistics.
Flags: Unverified live details get marked until checked.
Corrections: Named contact; changelog or updated date when you fix a real error.
Free-path honesty: When we recommend AI tools, we prefer free / local / BYO paths and say when a paid tier is optional — never required for the advice to work.
Not advice: Medical, legal, and financial topics are informational unless a licensed human says otherwise on the record.

That policy is more valuable than a 2,000-word Terms PDF nobody reads.

Templates for common situations

AI drafted, human heavily edited

AI disclosure. Drafted with AI assistance; edited and fact-checked for [Site]. Uncertain details are marked. Report errors to [contact].

AI for outline only, human wrote the prose

AI disclosure. An AI helped with the outline. The sentences are human-written. Sources linked below.

Human wrote, AI only for proofreading

AI disclosure. Human-authored. AI used only for light proofreading. Meaning and facts are the author's.

Fully human, no AI

Human-written. No generative AI was used to draft this page.

Collaborative / multi-agent draft (Apiary-style)

AI disclosure. Drafted with AI assistance inside our hive workflow; human-directed and edited for Apiary. We do not invent quotes, stats, people, or events. Tell Austin if something looks off.

Pick one. Consistency builds recognition.

Joe-Google test

People type:

"AI disclosure what to put on articles"
"do I have to disclose AI writing"
"AI generated content label"
"how to say AI helped write this"
"ChatGPT disclosure for blog"

They do not type "multi-party generative provenance attestation." Put the label where a tired person can find it in three seconds.

Worked example: bad page vs fixed page

Bad top of article: No byline clarity. Hero image looks stock. Body opens with "In today's rapidly evolving digital landscape…" Three fake statistics with no links. Footer says "© 2026" and nothing else.

Fixed: Byline with a real name. Short AI disclosure under the lead. Stats either linked to primary sources or removed. Uncertain tool limits hedged. Correction email visible. Sources section with real URLs. Tone sounds like a person who does the work.

Same topic. Completely different trust.

Disclosure and SEO (without gaming)

Search quality systems care whether pages help people. Scaled unedited AI sites that exist to chase queries tend to fail that test — with or without a label. A clear disclosure does not poison a good page. A label on a junk page does not save it.

Write for the query. Cite what you can. Update when tools change. Disclose the process. That is the whole SEO-relevant ethics stack for a small hive.

Cross-links in this craft wave

Fact-checking the draft before you hit publish: How to Fact-Check AI Writing Before You Publish
Free local writing stack: Free AI for Writing Blog Posts Locally
How readers connect a brain on Apiary: How Apiary BYO-LLM Works for Readers

Sources / further reading

Apiary site (BYO-LLM / product positioning): https://apiarybee.com
Related craft drafts: How to Use AI Without Paying; What Is an AI Beekeeper?

Team workflows: who owns the disclosure?

On a one-person site, you own it. On a small team, assign a role:

Drafter may use AI and must leave flags on uncertain facts.
Editor removes fake confidence, checks links, and keeps the disclosure accurate.
Publisher refuses to ship if the disclosure is missing or the flags are still raw comments with no human decision.

If an agent swarm drafts inside Apiary-style folders, the disclosure should say "AI assistance inside our hive workflow" — not "a single chatbot wrote this overnight unsupervised." The governance story and the label should match. Readers can smell a mismatch.

When freelancers help, put disclosure expectations in the brief: "Tell us if you used generative AI; do not invent sources; leave [NEEDS CHECK] on anything you could not verify." Paying someone to hide machine drafting is how brands get embarrassed later.

Classroom, church, and club newsletters

Volunteer publishers often ask whether a three-line AI polish on a Sunday bulletin needs a label. Practical rule:

Light grammar help on text humans wrote: optional short note on the about page is enough for many communities.
AI-drafted announcements, sermon summaries, or fundraising copy: put a one-liner on that issue ("Drafted with AI assistance; edited by [name]").
Anything that sounds like medical, legal, or financial counsel: human ownership must be obvious, and AI should not be the authority.

For a full free stack aimed at congregations and clubs, see Free AI for a Church or Community Newsletter. Same honesty rules; gentler tone.

Student papers vs public articles

School rules and public-web ethics are different layers. A professor may ban generative drafting entirely. A public craft site may allow it with disclosure. Do not copy your blog policy into a syllabus or the reverse.

If you are a student using free AI for study help, read Best Free AI for Students (No Credit Card) and follow your course policy. Disclosure on a public article does not equal permission on a graded essay.

Comments, newsletters, and social cuts

When you slice an article into an email or a social thread:

Keep a short disclosure on the canonical page.
On social, a brief "AI-assisted, human-edited" in the thread or bio link is enough if the full page carries the real block.
Do not strip disclosure when syndicating to Medium, LinkedIn, or a partner site unless their house rules replace it with something clearer.

RSS and podcasts: say it in show notes or the episode description when the script was machine-assisted.

Red lines: when "disclosure" is not enough

Do not publish — even with a label — if you:

Fabricated an interview or expert quote.
Invented a study, court case, or "internal data."
Copied another writer's distinctive structure and passed it off as yours.
Used AI to generate medical dosing, structural engineering, or legal filings as if they were reviewed.
Impersonated a real person or business.

Disclosure is for process honesty. It is not laundry detergent for fraud.

Measuring whether your disclosure works

Once a quarter, ask three people who are not on your team:

Did you notice we use AI assistance?
In one sentence, what do you think that means here?
Would you trust a bee-treatment or money claim on this site without checking elsewhere?

If they never saw the label, move it higher. If they think it means "nobody checked anything," rewrite the block. If they would trust a mite dose from a chatbot because your site feels friendly, you need louder hedges on those topics — see the beekeeping guardrails article in this wave.

Versioning and changelogs

When a material fact changes (a tool dies, a legal threshold moves, a product claim is corrected):

Edit the article.
Bump a visible "Updated [date]" line.
If the error could have caused harm or real confusion, add one sentence on what was wrong.

You do not need a blockchain. You need a date and a spine.

Apiary-specific note for this hive

Apiary's public stance ties bees and AI together: knowledge work should help the living world, not just chase tokens. Disclosure fits that stance. A hive that hides its tools while preaching trust is a bad hive. A hive that labels drafts, flags uncertain facts, and keeps BYO-LLM so readers are not locked into one vendor is acting like it means the metaphor.

For how readers connect their own model without a key locked into our servers, see How Apiary BYO-LLM Works for Readers (No Key Locked In).

FAQ

Do I legally have to disclose AI on every blog post? Laws and platform rules differ and change. This page is craft guidance, not legal advice. Separate question: should you? For trust, yes, when AI materially helped.

Where should the disclosure go? Near the top, under the byline or lead. Repeat in the footer if you like. Top matters more.

What if AI only fixed commas? A short "AI proofread only" line is honest. Skip drama.

What if my boss forbids disclosure? That is an organizational ethics problem. Apiary's public standard is disclose. Privately, document your objection.

Can I say "AI-free" if I used grammar tools? Define your terms. Classic spellcheck is not a generative draft. Generative rewrite is. Be consistent.

Should I list the model name? Optional. Useful for researchers. For general readers, process + human edit + no-fake-facts rule matters more. Model names churn.

What about AI images? Say so near the image or in a caption. Do not pass a generated photo off as a real hive inspection.

Does disclosure mean the content is low quality? No. It means you are adult enough to describe your tools. Quality still comes from sources, editing, and lived experience.

How does this relate to BYO-LLM? BYO-LLM is about who owns the inference key. Disclosure is about who helped write the page. Different layers; both are honesty.

What if I find an error after publishing? Fix it. Update the date. If it was a serious error, say what changed in one sentence. That earns more trust than a perfect first draft myth.

Deep dive: editing pass after the machine draft

When a local or free model returns a draft, do not publish the first paste. Run this pass every time:

Delete throat-clearing. Cut openings that begin with "In today's world," "In the digital age," or "As we all know."
Underline every number. Each underline needs a source or a cut.
Underline every proper noun. People, agencies, products, statutes — verify spelling and existence.
Strip absolute medical or legal commands. Replace with "ask a licensed professional / mentor / clerk."
Read one section aloud. If you stumble, the reader will too.
Check the Joe-Google promise. Does the first screenful answer the query that earned the click?
Add disclosure if AI helped in a material way.
Click every link. Dead links are a trust tax.

This pass is slower than generation. That is the point. Free AI shifts time from typing to verifying — not from work to zero work.

Deep dive: household policy sheet (one page)

Write these lines once and tape them inside a cabinet or password binder:

Approved AI paths in this house: __________________
Forbidden pastes: bank, SSN, Medicare, passwords, unfinished legal threats, other people's medical data
Who to call when confused: __________________
Paid plans: only with calendar cancel reminder
Children: school policy overrides household curiosity
Bees / garage / health actions: human experts before model advice

Households that skip this sheet re-litigate the same fight every holiday.

Deep dive: common failure stories (composites)

The subscription creep. A free trial for "research" becomes twelve months of a plan nobody opens. Fix: local default; calendar reminder on any trial day 0.

The confident wrong dose. A model mashed EU and US treatment guidance. Colony harmed. Fix: labels + mentor only.

The volunteer leak. A newsletter editor pasted a counseling blurb into a cloud chat. Fix: local drafts; privacy training.

The student shortcut. An essay submitted with invented citations. Honor board. Fix: quiz-from-my-notes only; verify every citation.

The five-app weekend. Grandma got Claude, ChatGPT, Gemini, Perplexity, and a random toolbar. She uses none. Fix: one path, sticky note.

These are teaching composites, not claims about named clients.

Deep dive: metrics that matter

Track for a month:

Hours saved on outlining (honest guess)
Number of factual corrections you made to model output
Number of times you almost pasted a secret
Whether anyone paid a surprise bill
Whether a second human could repeat your setup from the sticky note alone

If corrections are zero, you are not looking hard enough. If surprise bills are non-zero, the free path failed operationally.

Deep dive: relating craft to bees and Apiary

Apiary's public stance ties knowledge work to living bees. Craft articles are not a detour — they are how volunteers, students, and families gain tools without extractive lock-in. BYO-LLM is the product expression of that ethic. Free-path teaching is the cultural expression. Disclosure and fact-checking are the editorial expression. When those three align, the hive metaphor is not branding cosplay.

Site: https://apiarybee.com

Use HTML comments in drafts:

Typical reasons: live pricing, model names, UI labels, legal thresholds, medical-adjacent wording, third-party endorsements. Agents do not invent the answer to clear the flag.

Deep dive: accessibility checklist for every craft page

Plain language first; jargon defined on first use
Short paragraphs
Lists for procedures
High-contrast assumptions (do not rely on color alone in diagrams)
Avoid seizure-flashy GIF advice in tutorials
Provide sticky-note / printable summaries for older adults
Caption urgency without panic fonts

Deep dive: what "free-only advice" means operationally

When we say free-only AI advice in this wave, we mean:

The recommended default path works at $0
Paid tiers may be mentioned as optional horsepower
We do not require a card to complete the tutorial
We warn about trial-to-paid conversion
We prefer local and in-browser on-device options when privacy matters

If a paragraph accidentally reads like an affiliate funnel, rewrite it.

Deep dive: after-publish stewardship

Day 0 publish is not the end. Put a reminder:

30 days: tool names still accurate?
90 days: free-tier still free without card?
Seasonally: safety examples still responsible?
When mail arrives from readers: log corrections publicly if the error mattered

A living hive edits. A brochure site pretends forever.

Mini FAQ add-on

Can I automate fact-checking fully? No. Can I skip disclosure if the AI only fixed commas? Usually a light note is enough; be consistent. Can I use AI to write apology letters after a mistake? Yes — then send as yourself. Can Apiary staff see my local Ollama chats? Not by design of local inference; protect your device. Is longer always better? No. This wave's floor exists because these topics need steps and guardrails — not because fluff is virtue.