A few weeks ago, a client passed along some feedback they’d gotten on a newsletter we wrote with them.
The reader was so impressed with the newsletter they wanted to know who’d written it.
So our client gave us the credit.
But what we reminded them of later is that, while we did the typing, they’d already done the talking. The ideas, phrases, concepts, and content of the piece was theirs.
We’d prompted them, yes. We’d asked them unique and specific questions to draw out their insights, yes.
But they were their insights.
So whenever a client’s newsletter, social post, or other piece of marketing content gets praised, we give our client the credit.
That’s one of the sharpest benefits of working this way, instead of relying on AI or outsourcing to a ghostwriter: when it comes down to it, the content remains the client’s. Which means it remains an authentic demonstration of their value, not a demonstration of a tool’s or a vendor’s ability to fabricate the appearance of it.
I bring this up because, if Claude is writing your content, pretty soon it might start taking all the credit, too.
Anthropic just announced that they’ll be embedding watermarking technology into all their model outputs (for new model releases, beginning in the EU), so Claude-generated content will be detectable from those new models.
There were already AI detectors out there, but absent true watermarking, they were basically only guessing. Soon, AI detection (or at least Claude detection) will be far more accurate. Though, Anthropic notes, not fully conclusive.
Oh, and they also add in this little tidbit:
“The output can carry a Claude mark even if the underlying ideas, text, or data originated from another source.”
So even if you’re only using Claude for proofreading, if you copy and paste its output, it’ll include the watermark.
So who gets credit for the content when everyone believes that Claude wrote it?
Now, it’s possible Anthropic rolls this back, like their last brand-damaging watermarking effort. But they’re saying this one was prompted by EU regulation, so they might actually stick with it.
If you’ve done any AI-assisted development work using Claude, you already know that, by default, it appends a “Co-Authored-By: Claude” trailer to all commit messages.
You have to specifically tell it not to if you don’t want to advertise which tool you used to everyone who looks at your version history. Which is, at a minimum, pretty annoying.
And I think a lot of marketers, entrepreneurs, and advisors are going to be pretty annoyed that their posts will be watermarked and credited to Claude.
So why not just do it themselves?
I think the customary answer has become, “When I use AI, I can do so much more marketing,” but is that even good?
Have you heard the App Store stats? Because of the popularity of AI development tools, submissions to Apple’s App Store have roughly doubled this year.
That’s a lot more apps. So that means a lot more developers making a lot more money making software, right?
Right?
Right...?
No. App Store downloads are only up 2%!
It’s all just slop, and no one’s buying it.
That’s happening with marketing content, too. It’s impossible to know for sure how much more content is on the internet and social media now because of AI, but I think we can all safely assume it’s a lot more.
But is anyone buying more? Are they getting better advice? Or are they just wading through more noise to get to some nugget of value?
Slop content tells the prospect that, while the marketer didn’t invest any energy, thought, or experience into their content, the prospect should invest their energy and time into consuming it.
So when Claude takes the credit, what else will it be taking away? Will it be taking away some of the appearance of expertise, or the appearance of authenticity?
Or will it take away the appearance of physicality, the costly signal that says we’re invested in our business for the long haul?
See, when you give all of that away to AI, you give away more than time, effort, and energy.
You think you’re giving it a task, but what you’re really handing over is your voice.
Which means you’re handing over your ability to demonstrate your value to a machine that’s now demonstrating its value.
I know how tempting it is, how appealing it is, how exciting it is to have Claude take a burden off your mind and a task off your list. I know what a relief it can be to be able to get more done with the assistance of AI.
Most of my most-used apps on my phone I built using Claude, and they’ve helped me do incredible and exciting work. But there’s a limit, a line, past which I refuse to go.
Because some tasks are ours alone, and they cannot be given away without giving away something we were always meant to keep, because it’s what our clients are paying for:
Our expertise, as delivered through our insights, which demonstrate our abilities.
As I like to say, every decision is a marketing decision, including the decision to make content with AI, or to make content with your own mind. Your own experiences, your own real and embodied intelligence.
Because, at the end of the day, the way to demonstrate value is to deliver it. Not to fake it or fabricate it.
That starts by creating something new, novel, and something only you can do.
So that when someone needs the sort of thing you sell, they can trust you to deliver it to them.
Because you’re not faking it, you’re actually, truly, making it.
(If you’re worried that this change to Claude is going to degrade or detract from your marketing content, please reach out by replying to this email. I’ll show you how you can convert your current process into something uniquely yours.)
Kelford Inc. are content marketers for expert advisors. If you’re struggling with what to say and where to say it to attract your ideal clients, we’ll show you the way.