← Back to blog

AI & Content · November 15, 2026 · 8 min read

Do You Have to Disclose AI-Written Blog Posts?

Four parties can ask how a post was written, and only two of them want a label. Here is where each answer lives, the disclosure line that costs nothing, and three cases to assume yes.

By FluxWriter

Do You Have to Disclose AI-Written Blog Posts?

Do you have to disclose AI written content? Four different parties can ask that question and they do not give the same answer, which is why it feels unresolvable when you treat it as one question. Search engines, regulators, the platform you publish on and your own readers each care about something different, and a label only satisfies two of them. This separates the four, gives the disclosure line that costs nothing, and names the three cases where you should assume the answer is yes.


Four Questions Are Hiding Inside One

The query is short but it carries four separate problems. Is there a rule somewhere obliging me to label this. Does Google treat the page differently if I do. Do the terms of the site I am publishing on ask me. Will my readers think less of me. Nothing forces those four to line up, and most of the confusion online comes from an answer to one being quoted as the answer to all four.

The order of risk is inverted from how it feels. The part people worry about most is the law, and for an ordinary how-to post on a plumbing company's own site that is usually the part least likely to be in play, though whether it applies to you is not something to settle from a blog post. The part most likely to bind you is the checkbox you clicked past when uploading to a marketplace, which is a direct question you have already answered.

One more separation is worth making first. Disclosure and accuracy are different problems. A note saying a draft was AI-assisted does not repair a wrong price, an invented statistic or a testimonial from a customer who does not exist. Most obligations that genuinely exist in commercial publishing are about not misleading people, and they do not care which tool typed the sentence.

Google Asks For Quality, Not A Byline

Google's published guidance on AI-generated content is consistent on one point: its focus is whether the result is helpful and original, not how it was produced. It has not asked publishers to add an AI label for search purposes, and nothing in its public documentation treats such a label as a ranking factor in either direction.

What that guidance does say is that content produced primarily to manipulate search rankings is against its spam policies, which has always applied to human writing too. A page churned out at scale to catch a query, with nothing in it a reader could not get faster elsewhere, is the target. The production method is evidence, not the offence.

So do not add a disclosure hoping for credit, and do not withhold one fearing a penalty. Where process shows up in search performance it does so indirectly, through thin, unedited output that loses on merit to pages a person thought about. That is a quality problem, and no label fixes it.

The Legal Question Is The One Not To Settle Here

Two families of rules come up. The first is advertising and consumer-protection rules, generally concerned with whether a claim misleads and whether an endorsement is genuine. The second is the newer AI-specific transparency rules, the EU AI Act being the one people name. Whether either touches what you publish, in the markets you sell to, is a question for someone qualified in your jurisdiction. It is not something to conclude from this page.

What is useful here is knowing which questions to put to that person. Does what I publish count as advertising or an endorsement. Am I making claims about testing, results or personal experience that no person actually had. Does my sector have a regulator with its own content rules. Asked once, those questions are cheap compared with finding out afterwards.

If you work in health, finance, law, insurance or anything else supervised, assume the bar is different and ask before you publish rather than after. In those fields the answer is often yes for reasons that have nothing to do with AI, because the rules already govern who may give the advice and how it must be presented.

Platform Terms Are Where A Real Rule Usually Turns Up

If something obliges you to declare AI involvement today, it is far more likely to be a terms-of-service document than a statute. Self-publishing marketplaces have added a declaration step at upload. Several large social platforms ask uploaders to flag synthetic or altered media and attach labels of their own. Some writing platforms require a visible note, and guest-post editors increasingly put the question in the brief.

Those terms change often enough that any summary goes stale, this one included, so read the current wording where you actually publish. Search each set of terms for "AI" and read whatever surrounds it. That takes roughly two minutes per platform.

Where a platform does ask, treat it as what it is. A box asking whether content is AI-generated is not an invitation to a position on authorship. It is a declaration, and the cost of a false one is the account or the relationship. On your own domain there is no upload box to answer, but your host, your CMS and any ad or affiliate programme still come with terms of their own, and whatever governs your sector does not stop at your own domain name.

Counting the people you work for, five groups can ask about your process, and they are asking for different things. This is where each answer actually lives.

Who is asking What they actually want Where the answer lives How often it changes
Search engines Pages worth ranking, not a production credit Google's published guidance on AI-generated content Rarely
Regulators Claims that do not mislead Someone qualified in your jurisdiction Slowly, and by market
Platforms and marketplaces An honest answer to the question they asked The current terms of that specific platform Often
Clients and employers Whatever you agreed to Your contract, or their internal policy Per client
Readers To know who stands behind the page A line you write yourself Never, once written

Readers Punish Being Caught, Not Being Told

The trust question is the one with money attached, and it is not symmetrical. A plain note saying a draft was AI-assisted and edited by a named person rarely costs a reader's confidence. Being found out after implying that a person wrote every word costs considerably more, because the discovery reframes everything else on the site. The reader stops asking about that post and starts asking what else was not what it looked like.

What readers object to is rarely the tool. It is the false first person. "We tested fourteen grinders over three months" when nobody tested anything is not a disclosure failure, it is a claim that is not true, and it would be just as untrue typed by a human. The same goes for an author bio implying a qualification nobody holds.

A rough rule that survives most edge cases: if deleting every first-person experience claim would not change what a reader believes, the tool used is beside the point. If deleting them would gut the piece, the tool is the least of your problems, because those claims need a real person behind them.

Three Cases Where You Should Assume Yes

The first is when a platform asks you directly: a checkbox at upload, a field in a submission form, a line in an editor's brief. There is no judgement call in it. The question has been asked, and the only decision left is whether to answer it accurately.

The second is when you publish under someone else's name or on a client's behalf. What you are permitted to do there is set by your agreement rather than any general rule, so check what the contract says about how work is produced and raise it with the client if it says nothing. Plenty of clients do not mind. Some have an internal policy that does, and finding that out at the invoice stage is expensive.

The third is when the content leans on lived experience or professional judgement: first-person reviews, case studies, before-and-after results, guidance in any field where the reader is trusting a specific person's expertise. Here the disclosure is barely about AI. The reader needs to know who checked this and what they did, which is worth stating whether or not a model was involved.

The Disclosure Line That Costs Nothing

One sentence at the foot of the post does the whole job: "Drafted with AI assistance, reviewed and fact-checked by [name] on [date]." Eleven words, two of them placeholders. Put it in the post template rather than in individual posts and it covers everything already published and everything you publish next, at a cost of one edit.

Back it with a short editorial policy page linked from the footer. Say which tools are used, who reviews before publication, what gets verified, and how to report an error, with a real contact address. Roughly 300 words, half an hour once, and it answers the question before anyone has to ask it.

Three things to avoid. Do not put the line at the top, where it becomes the first thing a reader sees and the least interesting fact about the page. Do not write it apologetically. And do not leave it unattributed, because a note with no name reads as nobody here being accountable.

FAQ

Will a disclosure line hurt my Google rankings?

Nothing in Google's public guidance treats an AI disclosure as a ranking factor either way, so the reasonable assumption is that it is neutral. What moves performance is whether the page answers the query better than the ones above it. Judge the page on that, and treat the line as a reader-facing decision.

Do I still need to disclose if a person edits the draft heavily?

It depends on who is asking. If a platform puts a direct question in front of you at upload, answer it accurately regardless of how much editing happened afterwards. For readers, a line naming the person who reviewed and checked the piece is more useful than any attempt to measure the split between machine and human.

Can I use AI on client work without telling the client?

That is a contract question rather than a general one, and the answer lives in your agreement. Check what it says about tools, subcontracting and originality, and ask the client directly if it is silent on all three. Asking once is cheaper than being asked later.

The Practical Takeaway

Spend an hour this week on the mechanical part and leave the philosophy alone. List every place you publish, your own domain included, and note whose terms cover each one. Read the AI clause in each, answer any declaration honestly, and check your client agreements for anything covering how work is produced. Then add one line to your post template and write the editorial policy page it links to. Start by opening the terms of the one platform you publish on that is not yours and searching that document for "AI".



← All posts