The Case Against Letting AI Write Your Posts For You

The honest position on AI writing your posts is not “never” and not “always” — it is a line:

AI may write your caption, your hashtags, your promotional blurb. That is your content, and automating it is what makes posting sustainable.

AI may never write your customer’s words — a testimonial is theirs, verbatim. And it may never write a free-text post you mean sincerely — the personal note, the heartfelt thank-you, the thing that only matters because you wrote it.

The test is one question: whose sentence is this, really?

This is the Pen Rule as it applies to AI, and it resolves the whole “should I use AI or not” anxiety. You do not choose between all-AI and all-human. You put each piece of the post on the right side of one line.

Where AI belongs

For a busy owner, AI writing the routine stuff is not a compromise — it is what lets the posting happen at all.

• Captions — written in your voice, automated. Your content, your responsibility, fine to delegate.
• Hashtags — the fixed local set, applied automatically.
• Promotional copy — the “20% off colour in September” post. Yours.
• Transcription and subtitles — auto-generated, then checked for errors only.
• Distribution — publishing everywhere at once.

All of that is your content or mechanical work. Automating it removes the friction that kills the habit. There is no authenticity problem, because none of it claims to be someone else’s genuine words.

Where AI stops, absolutely

Two things AI must never write, and they are the two that carry all the value:

A customer’s testimonial. Their thirty seconds are theirs, verbatim — the “ehm”, the false start, the phrasing. AI may transcribe it (fixing mishearings), never rewrite it. The moment AI “improves” a testimonial, it becomes a fabrication, and a testimonial that reads better than the customer speaks is a fake one. This is not negotiable, and since 2024 a fabricated testimonial-or-review is also illegal.

A free-text post you mean sincerely. This is the subtler one. When you write a personal message — a heartfelt thank-you to your customers, a note about a bereavement, a genuine “we’re so proud of the team today” — its entire value is that you wrote it. Hand that to AI and it becomes hollow, because the sincerity was the point. Some posts matter precisely because a human meant them, and AI cannot mean anything.

The free-text distinction is the interesting one

Most people grasp “don’t let AI fake a testimonial.” Fewer grasp the second half, and it is the more subtle product truth.

There is a difference between a promotional caption (where AI is welcome — the content is functional, nobody needs it to be heartfelt) and a sincere free-text post (where AI is poison — the content is meant, and meaning cannot be automated).

“Balayage, £90, book Thursday” — let AI write it, who cares. “To the regular who’s been coming for twenty years and moved away today — we’ll miss you” — write that yourself, because if AI writes it, it is a lie about your own feelings. The value of the sincere post is the sincerity, and AI has none to offer.

So the free-text you write from the heart is published verbatim, exactly as you wrote it, because it is yours the way a testimonial is the customer’s. This is the Pen Rule pointing back at the merchant: we never moderate or rewrite the words you meant. Human versus automated is a line, not a switch.

The test that resolves every case

You never have to agonise over “should AI write this?” if you ask one question: whose sentence is this, really?

• The caption? Yours, functionally. AI may write it.
• The customer’s testimonial? Theirs. AI may not.
• The hashtags? Yours, mechanical. Automate.
• The heartfelt note? Yours, sincerely. Write it yourself.
• The subtitles of what she said? Hers. Fix errors, never rewrite.
• The promo blurb? Yours. Automate.

If the sentence is genuinely yours and functional, automate it freely. If it belongs to a customer, or it is something you mean, it is off-limits to AI. The line is not about AI being good or bad — it is about whether the words are someone’s genuine expression, and genuine expression cannot be delegated.

Why this matters commercially, not just ethically

This is not just principle. It is why the product works.

A business is not made more trustworthy by better-sounding testimonials, or by a heartfelt post an algorithm wrote. It is made trustworthy by real words — the customer’s, and yours-when-you-mean-them. Every time AI crosses the line and writes what should have been genuine, it destroys the exact thing that was persuasive, and a customer can feel the hollowness even if they can’t name it.

So the AI that writes your captions and spares your evenings is a good tool. The AI that writes your testimonials and your sincere posts is quietly dismantling your credibility. Same technology, opposite outcomes, decided entirely by which side of the line you point it at.

But AI writes it better than I do

Say you read the AI version of your thank-you note and it does sound smoother. That is exactly the trap. For a promo, smoother is better — nobody needs “£90, book Thursday” to have any feeling in it. For a post you mean, smoother is worse, because the small awkwardness is the signal. The slightly clumsy line, the sentence that trails off — those are the hesitations that prove a person wrote it, the same way a customer’s “ehm” proves the testimonial is real. Polish it and you sand off the one thing that made anyone believe you. The goal was never a good post. It was your post. Writing your own social posts is not a skill you are failing at — for the sincere ones, it is the whole point.

A florist, a Tuesday, two posts

Picture a florist with two things to post on the same morning. One: Mother’s Day pre-orders are open, order by Friday. Two: a customer who bought flowers from her every week for years has died, and the family has asked her to do the funeral arrangements.

The first post, hand to AI without a second thought — it is a functional notice, and AI will phrase it cleanly while she serves the queue. The second, she writes herself, in three plain sentences, one of which doesn’t quite land grammatically. That imperfect one is the post her regulars reply to, because they can tell she stopped to write it. Had AI produced a tidy paragraph about cherished memories, they would have scrolled past. Same shop, same morning — the line ran straight between the two posts, and the one question told her which was which.

Use AI for the functional, keep the meant human

Let AI write your captions, your hashtags, your promos, your distribution — the functional, yours-anyway content that was eating your evenings.

Keep two things human, always: the customer’s words, and anything you actually mean. Because those are the parts whose entire value is that they are genuine — and genuine is the one thing a machine cannot produce.

Whose sentence is this, really? Answer that, and you always know where AI stops.

The broader line — what to automate and what to keep human — is the piece this rests on.