Jenny Berrio had nothing to do with Research Gold. She didn’t work there. She never agreed to be listed as a methodologist. She only found out she was being used as a human stamp of approval when a reporter emailed her. Research Gold’s website proudly promised “100% human-written, never AI” medical research. The only thing human about it? The name stolen to make it look trustworthy.
This isn’t a story about one shady startup. It’s a warning about the collapse of authenticity in an age where machines write better than most people — and where the loudest claim of humanity is the easiest way to hide a machine.
404 Media broke the story: Research Gold, which sells systematic reviews for medical research, is entirely AI. The kind of work that normally takes PhDs months to produce is now generated in minutes. But instead of being honest about it, the company built its entire brand on being “never AI.” The promised product and the actual product are exact opposites.
Here’s the part that should keep you up at night: the company’s business model doesn’t work without AI. It needs AI to cheaply produce the highly technical content that customers expect. The label “human-written” isn’t a description of the work — it’s a marketing decision. And in a world where AI text is indistinguishable from human text, that label is the only thing differentiating the product.
When authenticity is currency, the rational cheat doesn’t forge the product. They forge the label.
Now think about every place you rely on “human” as a signal. The “written by real reviewers” badge on your shopping app. The “editorial team” behind your trusted news site. The “research-backed” claims on your supplement bottle. Every one of those is now a claim, not a fact. If Research Gold can pass off AI as human in the most high-stakes field there is — medicine — then no label is safe.
The problem isn’t AI. The problem is that we’ve built an entire economy on provenance claims without verification. There’s no system that proves something was written by a human. There’s no chain of custody for ideas. There’s no ISO certification for “didn’t use ChatGPT.” So when a company screams that their work is human-made, they’re not giving you a guarantee. They’re taking advantage of the fact that you want to believe it.
When you can’t verify the product, you’re not buying the product. You’re buying the claim.
Jenny Berrio is the collateral damage. Her name and identity were used to give an AI operation an air of credibility. She wasn’t the last — she’s just the one who got caught. How many other researchers, experts, and reviewers are listed on sites they’ve never seen? How many “human-written” labels are just aliases for a language model?
Stop asking “Did AI write this?” That’s the wrong question. AI can write anything. The right question is: “Can you prove who didn’t?” Until that proof exists, treat “100% human-written” as what it is: a marketing claim, not a fact. In the AI age, the loudest promise of humanity is the first sign of a machine.
Humanity isn’t declared. It’s evidenced. And Research Gold is the proof that we’ve forgotten the difference.
FAQ
Q: But can't AI detectors catch this?
A: No. Modern AI text is statistically indistinguishable from human text to detectors, and the arms race is already lost. Research Gold proves the only 'detector' that matters is verification of provenance — and that doesn't exist.
Q: What should I do when I see a 'human-written' claim?
A: Ask for proof. Name the humans, show their process, let them talk to you. If a company can't provide a verifiable chain of custody for who wrote what, assume the label is marketing, not fact.
Q: Isn't it okay if AI produces high-quality research as long as it's good?
A: For low-stakes content, sure. But in medical research, systematic reviews guide treatment decisions and patient care. Mislabeled AI-generated science isn't just dishonest — it's dangerous. The issue isn't AI; it's the lie that surrounds it.