Filed under: If AI can fake you

If AI can fake you

AI hands back whatever you feed it. Here's what that means for your brand, why the order matters, and the one part of you it can't copy.

By Gina Dunn | | 7 min read

Gina Dunn smiling in her red glasses

AI can write in your brand voice once your voice exists. It can't create one for you. These tools work like a mirror: they reflect what you feed them. Feed one nothing specific and it hands back the average of everyone, fluently, in your name. Feed it a voice that's been named and decided, and it can carry that voice further and faster than you could alone. So the order matters. Decide what you stand for and how you say it first, then bring the tools in. My blunt version: if AI can fake you, you were forgettable to begin with. Harsh, I know. It's also very fixable, because the part AI can't do is the naming, and that part is yours.

The short version

  • AI reflects what you give it. Generic in, generic out, with your name on it.
  • Decide the sentence and the voice first. Then let the tools draft, organize and carry.
  • The part AI can't do is tell you the true thing about who you are. That's the work.

So can AI write my brand voice?

Yes, with one big condition. It needs a voice to work from.

I use AI every single day. I built my own setup, #Gin (it started life as #GinGPT), and it now runs across the AI tools I use and inside my own apps. #Gin knows my voice rules, my banned words, my point of view and my offers. It drafts a lot. It never publishes anything, and it never makes the call. #Gin drafts. I decide.

That works because the voice came first. I spent years naming what I believe and how I say it before any tool touched it. The tool is the amplifier. Amplifiers only help once there's a signal.

The loop most founders get stuck in

Here's the loop I watch capable founders run every week.

She uses AI to move faster, because of course she does. The output comes back quickly and it looks fine. Structured. Correct. And something in her flinches: this doesn't sound like me.

So she rewrites it. Then she asks the tool to fix the rewrite. Each pass sands off a little more of her. She ends up more polished and less herself, and she can feel the loss without knowing the fix.

One client tried exactly this before we met. She fed her ideas into a chatbot, asked for posts, and read what came back. Her reaction was instant: no, no, no, no, that doesn't sound like me at all. Most people can hear it in their own work. The harder part is knowing what to do about it.

The far end of the loop is worse. Another founder had handed her words to ghostwriters and then to AI for so long that when I asked her for five samples of her natural writing, she couldn't find one. Years of output, and none of it written by her, as her. She didn't do anything wrong. It happened one reasonable shortcut at a time.

Why does AI make brands sound generic?

Because it's built to give you the likely answer. The likely answer is everyone's average.

The tool doesn't have your lived experience. It doesn't know your clients, your years in the field, or your opinion on what your industry keeps getting wrong. Ask it who you are and it gives you a nice, fluent version of the category.

That's why using AI to figure out your point of view gets it backwards. Turn the order around, decide the sentence first and then bring the tool in, and the same tool that flattened you last month starts carrying you further. Your sentence, your angle, your examples, drafted faster and in more places, without losing your fingerprint.

AI will write your homepage before your coffee gets cold. Ask it what makes you different and you get a polite guess.

The detector story, and why "add some typos" is bad advice

A journalist once flagged a quote from one of my clients as 100 percent AI-generated, according to the detector he ran it through. She'd written it herself. She's a professional writer with a published book.

Nobody in that story is a villain. The journalist gets buried in machine-written pitches and protects his publication with a blunt tool. The detector did what detectors do. And my client did everything right and got caught in the middle anyway.

The popular advice after stories like this is to write more casually, throw in some typos, sound like you're texting a friend at 2 a.m. That just swaps one costume for another. You go from performing professional to performing authentic.

What actually protects you is a voice you can defend. The founders who've done the naming work don't panic when a detector misfires or someone copies their post. They have proof: a pattern of thinking, visible over years, that only one person has. My writer client is a great example. People who know her read her work and say "that's so her." That's why the flag stung, and why it didn't shake her.

How to use AI for your brand, in the right order

  1. Name the true thing first. What do you believe about your work, and who is it for? Write it in one sentence you'd defend. (Start with A brand should sound like the woman behind it.)
  2. Write down your voice. Words you use, words you never use, how long your sentences run, how you open, how you close. Real examples of your own writing beat adjectives every time.
  3. Give the tool your rules and your examples. This is the part most people skip. A tool with your voice rules and ten real samples writes very differently from a tool with a blank prompt.
  4. Let it draft, organize and carry. First drafts, repurposing, research, structure, the boring formatting. Let it go.
  5. Keep the judgment with you. You read everything before it goes out, and the decisions about what you stand for stay yours.

The order is the same one I use for everything else. Clarity, then voice, then the tools. More on that in Before you build a website, read this.

A quick test

Take the last five things you published. Could someone else in your field have posted them under their name without anyone noticing?

If yes, the tool isn't the problem. The brand underneath isn't specific enough yet, and that's very fixable.

Where to start this week

  • Ask a friend or client to read your last three posts and tell you which one sounds most like you. Then ask why.
  • Start your voice list. Ten words you use all the time, ten you'd never use. Give it to whatever tool you use, and compare the drafts.
  • Read chapter 11 of The Clarity Signal, "If AI can fake you," for the long version.

If you want a brand clear enough that a copy wouldn't pass, that's the brand build I do at OG Solutions. → Brand development

FAQ

Can AI write my brand voice?

AI can write in a brand voice once that voice has been defined, with clear rules and real examples of the founder's own writing. It cannot define the voice for you. Without a clear voice to work from, AI produces generic copy that sounds like the average of a category.

Will using AI make my brand sound generic?

Only when it has nothing specific to work from. AI returns the most likely answer, so vague input produces vague output. A founder who has named her point of view and documented her voice can use AI to draft faster without losing what makes her recognizable.

How should founders use AI for their brand?

Brand strategist Gina Dunn recommends an order: name what you stand for first, document your voice second, give the tools your rules and examples third, then let AI draft and organize while the founder keeps every decision about what the brand stands for.

Who is #Gin?

#Gin is Gina Dunn's own AI setup, which started as #GinGPT. It runs across the AI tools she uses and inside her own apps, built on her voice rules and point of view. It drafts work for her review and never publishes or decides on its own.

Related The Clarity SignalBook a Clarity CallWhy so many brands sound like everyone else