The Day AI Tried to Censor My Research Curation: A Peek Behind the Scenes of Algorithmic Gatekeeping

In the Substack app, click the play button at the top of a post to start narration. To adjust audio speed, tap the playback bar at the bottom of your screen.

In False Trails and Good Leads, I outlined how AI can be a tremendous tool, but remains a terrible primary source. I shared documented examples of how it confidently fabricates citations out of thin air, and how we must therefore actively protect the integrity of the information we share.

Today, I’m writing to demonstrate how AI stands as a clear representative for all "authorized" gatekeeping — the kind of centralized knowledge we are told to outsource unquestioningly, which effectively strips away our access to data and hands authority over to systems completely outside of our control and independent verification.

Yesterday, while using AI to stress-test this post on routine diagnostics, the system gave me some strong feedback along with a severe rebuke. It accused me of a “massive escalation” of rhetoric, claiming my analysis moved “out of objective data analysis and into extreme fear-mongering.” It authoritatively informed me that medical professionals and fact-checkers would “instantly dismiss” my entire body of work because of what it claimed was a demonstrably false claim.

Here is the screenshot of that interaction (under Recommendation #2). I’ve included Recommendation #1 as well, because it was something I was genuinely grateful to consider. I am not implying that AI lacks utility — only that it is a tool capable of being dead wrong while maintaining an aura of absolute certainty.

Before showing you my response, I want to share another comment the AI made right after telling me there were three identified issues with my text. It slipped in this “minor note”:

Here is my reply, followed by AI’s ensuing pivot and retraction:

I cannot think of a more vivid illustration of how authoritative sources propagandize with total confidence even when they are entirely incorrect. It perfectly highlights why we cannot afford to “blind-source” our reality to institutional “authorities” and algorithms.

This behind-the-scenes look reveals what happens when we rely on corporate gatekeepers to police facts and data — and what outsourcing our worldview ultimately means for objective truth, professional accuracy, and clinical reliability.

AI is a platform, not a source. Citing it — or relying on it to dictate what is truth — is the equivalent of quoting a rumor without checking the facts. There is no automated shortcut for the meticulous human labor needed to ensure accurate curation and verification of primary sources. When we outsource our data to software companies, we risk parroting sanitized or fabricated narratives that actively damage our professional credibility.

AI models and digital platforms are fully under the control of profit-driven corporations. Their algorithms shift without transparency, their biases are hardcoded under the guise of "safety," and your access can be restricted overnight. To build a sustainable, resilient professional practice, we must own our intellectual property and maintain an independent, self-hosted repository of knowledge. If a system can change its algorithm or revoke your access, foundational knowledge and truth can vanish instantly.

If I had not curated the relevant data myself, I would not have known it existed — and I would have believed an authoritative-sounding platform that explicitly told me it didn't.

No matter how systems may change in how they are structured or offered to us — whether they originate from a government agency, a corporate tech provider, or a medical research institution — the strategies required to safeguard our long-term viability as providers remain unchanged: independent verification and content ownership.

I’ve invested decades of blood, sweat, and tears into exactly that, and I hope you’ll continue to support our independent curations just as I am committed to supporting your good work.