We didn’t see it coming.
A new study dropped on August 24, 2025, from Originality.ai — an AI detection tool — and the numbers are jaw-dropping.
Of 2,034 recently published religious books on Amazon’s KDP platform, 63% were flagged as “likely AI-generated.”
Not “maybe.” Not “could be.” Likely.
That’s a landslide.
Now, replace “religious books” with “crypto research reports.”
Think about it. The same incentives — low production cost, high volume, low reader scrutiny — apply to the crypto space. Whitepapers, market analyses, even project documentation. How much of that is AI-generated?
We don’t know. But we should.
Context: Why This Matters Now
I’ve been in this industry for 24 years. I’ve seen the ICO boom, the DeFi summer, the NFT frenzy. Each time, the hype cycle was driven by narrative — not just technology.
Now, AI is the new narrative accelerator.
Originality.ai’s study is a canary in the coal mine. It shows that AI-generated content has crossed the threshold from “experimental” to “dominant” in a specific vertical. The religious book category is just the first.
Crypto is next.
Why? Because the barrier to entry for crypto content is even lower. Anyone can write a Medium post, a tweet storm, or a protocol whitepaper. There’s no editorial gatekeeper. And the economic incentive? Massive. Fake news, pump-and-dump narratives, and FOMO-bait are cheap to produce.
Root: The problem is that AI detection tools are not reliable. Originality.ai itself admits its results are “probabilistic, not deterministic.” False positives exist. False negatives are rampant.
This is exactly the same trust issue we face in crypto oracles.
Core: The Technical Breakdown
Let’s dive into the numbers.
The study found that the AI-generated percentage varied by subcategory: magic/occult books hit 78%, while Hinduism and Taoism were lower but still above 50%.
Why? Because these niches have low knowledge density. Readers can’t easily verify facts. The same dynamic applies to emerging crypto sectors — like AI agent tokens or DePIN protocols.
But here’s the kicker: the detection tools are playing catch-up.
Based on my experience auditing smart contracts, I know that any tool that claims to detect “AI-generated text” with high accuracy is lying. The underlying models — GPT-4o, Claude 3.5 — are constantly evolving. The statistical signatures shift.
s Demo of Originality.ai’s tool might work on ancient GPT-3.5 text, but it’s blind to the latest outputs.
So the 63% figure is likely an undercount. The real number could be 70%, 80%, or even higher.
Now, transpose this to crypto. Imagine 63% of all new crypto research is AI-generated. How many of those “independent analyses” are actually just LLM hallucinations?
We don’t know. And that’s terrifying.
Contrarian: The Unreported Angle
Everyone is focused on the content quality problem. But the real story is about trust infrastructure.
In crypto, we solve trust with consensus mechanisms, cryptographic proofs, and on-chain verification. But content platforms like Amazon — and crypto media outlets — lack this. They rely on centralized detection tools that are fundamentally adversarial to the AI generation process.
That’s a losing battle.
The contrarian view: The solution isn’t better AI detection. It’s decentralized content provenance.
Imagine a protocol where every piece of content is timestamped and signed by the creator’s wallet. AI-generated content would have a different signature — or no signature at all. Readers could verify the origin on-chain.
The party doesn’t start until we build this.
Until then, we’re flying blind. The 63% number is just the tip of the iceberg. Below the surface, AI-generated content is flooding every platform — including crypto research.
And the worst part? The market is pricing it in. Investors are buying narratives, not facts.
Takeaway: What to Watch Next
Here’s my forward-looking judgment:
Within 12 months, we’ll see a major crypto project caught with an AI-generated whitepaper. The backlash will be massive.
But the real opportunity? Build the on-chain content verification layer.
We didn’t ask for this, but we have to deal with it.
The question is: will you be the one creating the noise, or the one cutting through it?
