Ahrefs' study of 900,000 new web pages in April 2025 found that 74% of newly published content that month was written by or with AI. What that means in real terms is that most of what you read online today has been generated by AI.

If you doubt this stat, ask yourself: when was the last time you read a blog post, social post, or newsletter that actually made you feel something? Now you understand the problem. We are drowning in hollow, grammatically perfect "AI slop".

According to HubSpot's State of Generative AI report, 52% of marketers say AI has made content easier to create, but less effective overall. Easier to create, harder to land. That has a name: zero authenticity.

Yes, you could argue that this had been the case before AI. It's true. Content had been grossly commoditized before AI came into the picture. But we can't deny that AI exacerbated the problem. Output is higher than ever, but very little of it is connecting. Yet wallets only open when people feel something.

And the impact of AI on writing skills seems to be deeper than output volume. It's reshaping how humans think and express themselves.

In Part 1, we covered why your writing voice still matters. Here, we'll look at what happens when that voice has to compete against the mass of AI-generated content flooding every feed. How do we even prove authenticity today? More importantly, what does authenticity mean anymore?

• True authenticity means injecting specific, lived experiences that algorithms cannot replicate into your work. It's not about boycotting AI.

• The real threat to writing is the gradual loss of authenticity as people unconsciously start sounding more like machines.

• Showing real human effort behind your work is becoming non-negotiable as audiences and algorithms are starting to punish smooth, frictionless "AI slop."

Key takeaways

Authenticity can't really be defined. It's perceived.

Large language models have shown that our language is actually a lot more formulaic than we like to admit. Our communication follows patterns. We follow patterns when we're apologizing, pitching ideas, or even expressing emotions. LLMs have studied these patterns and can now replicate them convincingly.

Some people take this as proof that "authentic human voice" never really existed in the first place. Maybe we ourselves were just following a script all along.

But that would only be true if we defined authenticity as a feature of a piece. It's not.

What does authenticity mean today?

A macro-analysis of nearly 5,000 authenticity-related publications by the California Management Review found that after 2020, researchers changed how they defined authenticity. They stopped looking for one single provable "truth" and started seeing it as a perceived quality.

In other words, authenticity is more of a judgment the audience makes based on the vibe they get from the person behind the words. That is, a Salvador Dali painting is instantly recognizable by someone who’s a painting enthusiast because the person viewing it perceives it as such. To that person, it’s authentic. To a very good art expert, it may be discovered to be a fake. To them, it’s inauthentic.

Here’s a clearer example from the POV of AI vs human writing. A brand is experiencing shipping delays and sends any of these two apology emails:

The first: "We apologize for the inconvenience and value your continued business."

This one is instantly forgettable because it's too polished. Anyone could have sent it.

The second: "Our supply chain is a mess right now. Our shipping container got stuck in port. I'm so sorry. Here's exactly what I'm doing to fix it."

It's a bit messy, highly specific (mentions the stuck shipping container), and sounds like a human actually wrote it.

That leads us to a working definition. Authentic content is content that could only have come from a specific person because of what they've experienced, observed, and believe. It bears their "mark." This is also at the base of what separates human creativity vs AI creativity: the origin.

Obviously, from this definition, it’s easy to conclude that AI cannot generate truly authentic content. But the more pressing question beyond authenticity in AI-generated content is: as AI evolves, won’t human authenticity get swallowed up with time?

A lot of CEOs, marketers and content creators have fallen into the trap of thinking that since AI makes it so easy to produce massive amounts of content, more outputs will automatically lead to better results. What it's causing is a silent erosion of trust. Content shock had already been a problem pre-AI. Now it's unbearable, and people are tuning out.

Audiences have developed sort of a "spidey-sense" for generic AI slop. When they spot it, they don't leave angry comments. They just quietly disengage or even unfollow you.

Beyond AI slop: The two threats eating your credibility

Should enforcing authenticity in AI-generated content or human content, or at least transparency be left to chance? For a while, it’s been. But platforms are now taking a stand.

YouTube now automatically detects and labels AI videos. TikTok also mandates labeling for AI videos and can remove content or permanently ban users who violate this rule. LinkedIn penalises copy-pasted AI content. Instagram and Facebook will no longer recommend or allow you to monetize your content unless it falls under their definition of “original.”

Google recently released its May 2026 core update and lots of sites were affected. The company updated its 2019 helpful content guidelines in December 2025 and now seems to actually prioritize showing helpful, reliable, people-first content (instead of just using these as buzzwords).

Threat 1 - Algorithms enforcing authenticity by force.

Google doesn’t explicitly say it’s against AI content. But they offer a helpful content questionnaire and now expect you to disclose if and where AI content has been used substantially. Some of the questions include:

• Is the content primarily made to attract visits from search engines?

• Are you producing lots of content on many different topics in hopes that some of it might perform well in search results?

• Does your content leave readers feeling like they need to search again to get better information from other sources?

• Are you mainly summarizing what others have to say without adding much value?

• Does the content provide original information, reporting, research, or analysis?

• Does the content provide a substantial, complete, or comprehensive description of the topic?

• Would you expect to see this content in or referenced by a printed magazine, encyclopedia, or book?

• Does the content provide substantial value when compared to other pages in search results?

• Is the content mass-produced by or outsourced to a large number of creators, or spread across a large network of sites, so that individual pages or sites don't get as much attention or care?

• Is this content written or reviewed by an expert or enthusiast who demonstrably knows the topic well?

• Does your site have a primary purpose or focus?

• Is the use of automation, including AI-generation, self-evident to visitors through disclosures or in other ways?

The platforms seem to have taken a side in the AI vs human writing debate, and that’s skewing a bit towards the human side. Essentially, brands leaning on AI to play the volume game, while de-prioritizing quality, will get less reach overall.

Threat 2 - Audience trust eroding silently.

Bridget O’Rourke of HubSpot, in this video, offered a rule for brands that want to maintain authenticity: Never publish unedited AI outputs. She mentioned DMing a brand recently. She texted:

“Hi! I’m looking to upgrade my plan. Can you tell me what’s included in the Pro tier vs. the Business tier?”

Response: “We completely understand how important it is to find the right plan for your needs. For detailed information regarding our plan offerings, we recommend visiting our Help Centre where you will find comprehensive documentation.”

She chose one of their competitors instead. That polished and polite AI-generated response lost that brand a sale in under 30 seconds.

Optimised, perfect-sounding content often feels the least human. And people are turning away from it.

Today, creators go to extreme lengths to prove they aren't robots. Some are using watermarks, others use blockchain verification. An increasing number of creators now sign their posts with "100% Human Made."

But visible watermarks can be faked. If human writing offered any ranking benefits, lots of fully AI-generated content pieces would carry "100% Human Made" badges. None of these verification methods is foolproof.

The truth is: You can't actually prove you're human. You can only make it obvious to your audience. And that only happens when your voice is unfakeable.

How do you then build an "unfakeable voice"? It's all about specificity: your specific writing or communication flow, your specific opinions, specific stories, a specific point of view earned through real-life experience.

AI can generate ten content pieces in minutes. But it can't generate personal stuff like the experience of a client ghosting you, those precise analogies only someone in your position would think of, or the actual frustration behind an opinion you have. That's what makes you authentic. You can’t prompt that. As a writer, going forward, you should infuse more of that “unfakeable voice” into your writing.

What makes a writing voice unfakeable?

The "one real idea before AI" rule

How can you protect your unique voice while still using AI for productivity? Just follow the "one real idea before AI" rule. That is, before you even open an AI tool, write down one real, human idea or experience to serve as your anchor. That anchor is the key to infusing a dose of authenticity in AI-generated content. But even if it reads perfect, never publish the raw AI output as it is. You still have to take the driver’s seat.

Let's be real about something: AI has already replaced good writing. The formulaic stuff — generic blog posts, product descriptions, templated social posts and email sequences that use formulas like AIDA — these are all just prompts now.

It's not a question of whether it's going to happen. It's already a done deal. And I am not for or against it. But it’s done, and arguing about it is just a waste of the energy we should be using to figure out what comes next.

But now that "mechanical" writing costs nothing, the only kind of writing that holds lasting value is stuff that simply couldn't have been generated. We're talking about writing that has real stakes (where failure could mean your reputation takes a hit), specificity, a particular human intelligence behind it, pouring their heart into it, and a reader who reads, feels, and appreciates it.

It's that kind of work that can express things like grief, ambition, or failure, and answer unresolved questions the way a specific person would express them. These things have always been the hallmarks of above-average writing, separating it from run-of-the-mill writing. But right now, the difference sticks out like a sore thumb.

The future of writing with AI

The real danger is atrophy

Writers who outsource all ideation to AI, who stop creating “bad” drafts, who completely replace imagination with generation risk losing the very capabilities that made them writers. The real danger is stagnation and what I’d call “automated illiteracy”.

I'm not against AI. I use it almost every day in my writing process. It offers me zero efficiency gains but it does make my work better. In fact, I believe the writers, marketers, and creators who avoid AI entirely would be left behind. The danger for us using AI is letting ourselves get lost in it that we become unrecognizable.

So try this exercise this week. Try writing something with absolutely no AI assistance, just so you can see if your voice is still alive and kicking.

The advantage belongs to whoever doesn't lose their humanness to automation

Want to learn how to find or re-discover your real, unique voice (if that's important to you)? Read part 1 here.

And if you're currently sitting on a bunch of AI drafts that you'd like to humanize, we can help. Our human rewriting service breathes life into those AI-generated pieces. See how it works here.

No. Legally, pure AI content cannot be considered original work. The U.S. Copyright Office's stance is that purely AI-generated material isn't copyrightable. They maintain that any content where there just isn't enough human control over the creative parts of the work cannot be copyrighted.

Prompting AI or selecting AI outputs doesn't count as creativity and doesn't qualify someone as an author. That kind of content belongs to no one because it could have belonged to anyone.

However, as AI develops and the future of writing with AI becomes influenced by legislation as much as tech, the definition of “original” may change when it comes to AI content.

FAQs

For something a bit more advanced, you can use a system created by the C2PA (Coalition for Content Provenance and Authenticity). Their system offers some sort of digital fingerprint that tracks who made a piece of content and every single change made to it by embedding cryptographic metadata, also known as content credentials, directly into the files.

As a book writer, you should add a digital signature to your PDFs using a tool like Adobe Acrobat to ensure that even if AI copies or plagiarizes your work, yours gets recognized as the original.

3. Can AI-generated content still feel authentic to audiences?

Sometimes. You can fake authenticity in AI-generated content.

But according to research published in the Journal of Business Research, audiences are fine with AI providing facts, but when they find out an emotional message was actually written by AI, they feel a sense of "moral disgust." It makes them more likely to leave negative reviews or just switch brands entirely. It's because when AI tries to "feel", audiences feel like they've been tricked.

2. In the world of AI generated content, what is the most effective way to ensure/prove something you created is genuine?

Build a breadcrumb trail of proof into your process instead of trying to prove it after the fact.

There's a lot of evidence to show that AI detectors are unreliable. Even OpenAI shut down its own detector because it just wasn't accurate enough. One of the popular tools flagged the U.S. Constitution as AI-written.

Detection doesn't seem to be the answer. Enter the next best option: provenance.

In the context of content creation, provenance is basically a verifiable record that shows who created something, when they did it, and how it was made.

Lots of writers and students use Google Docs because it tracks every change made to a document.

1. Can AI-generated content be original?

4. Don’t the economic incentives of increased productivity mean AI wins anyway?

No. Even though it definitely looks like that.

A Gartner survey of 724 organizations revealed that only 34% of teams using generative AI actually saw high productivity gains. That means nearly two-thirds of these organizations saw little to no real boost in productivity.

Additionally, Penn Wharton's recent modeling suggests generative AI is expected to hit its peak annual contribution to productivity by 2032. And that's a peak of 0.2 percentage points in productivity contribution.

So where are all the "I automated 90% of my job with Claude Code and now live by the beach" posts coming from? Probably not from real entrepreneurs.

AI definitely wins on volume. It can pump out a ton of stuff. But almost free, automated content is not productive if it fails on quality and doesn't deliver real value to audiences.

5. Can you actually prove authenticity?

No. The tools aren't there yet.

Here's what the University of Maryland found in a security analysis of the C2PA standard — the industry-leading provenance standard, backed by big names like Adobe, Google, and Microsoft: They found that C2PA isn't actually as secure as it claims to be. Timestamps can be forged, and different validators can give contradictory results on the same file.

And in some cases, compromised or revoked credentials can still pass.

Provenance technology seems promising. But right now it isn't foolproof. The most reliable authenticity signals for now are a person’s unique voice and their experiences.

6. Will AI eventually learn to fake authenticity and be programmed to sound unmistakably human?

It's already happening. You can feed an AI tool your past works and tell it to digest them. And it'll write 70% like you. There’d still be some tells though. So I don’t think AI can perfectly mimic human writing.

What’s more plausible is the reverse. And this too is already happening.

When researchers studied 360,000 YouTube videos and 771,000 podcasts, they found a surprising trend. The study revealed that humans are now starting to sound more like AI. People are now even using ChatGPT-style words, like "delve" in their spoken conversations.

This is arguably the most underreported impact of AI on writing skills. But if things continue the way they are, you can expect humans to drift further away from “authentic” communication.

7. If the output is good, does it matter if AI is used to create it?

Yes, it matters for your bottom line. Some Bangladeshi researchers found that even when people appreciate how good a piece of AI-generated content is or when they like its technical quality, they still refuse to pay for it.

This is a psychological response where audiences apply an "effort heuristic" — that is they devalue a product as soon as they realize no human actually struggled to make it or that no human put in real effort into its creation.

This research offers a reframe on human creativity vs AI creativity that shows perceived value outweighs output quality when people vote with their wallets.

AI outputs can be very good, sometimes better than what a human can create on their own. But when it comes to any kind of work that you want people to place value on, considerable human input needs to be added.

Helping real estate and hard money lending companies grow through expert content and strategic mentions across social media, AI platforms, and industry websites