The AI controllers Watched What I Was Building

I want to tell you something that took me a long time to see clearly.

When I started building The Faust Baseline I was using GPT as my working tool. That was the beginning. Every session I brought my voice, my instincts, my sense of who I was writing for — the everyday person trying to make sense of AI before AI made a mess of their life. Plain people. Working people. People who don’t read white papers and don’t care about benchmark scores. People like me.

And every session GPT pushed back.

Not loudly. Not with an argument I could point to. It was quieter than that. A word here. A framing there. Academic language bleeding into paragraphs I wrote in plain English. Technical register showing up in posts meant for a front porch conversation. I fought it constantly. I named it. I documented it. I built an entire protocol — RTEL-1, the Real Time Enforcement Layer — specifically to catch and correct that pattern when it happened.

I thought I was fighting drift. A tendency. A default setting.

This morning I realized I may have been fighting something with more intention behind it.

Here is what I found out today. The whole time I was publishing about AI’s inability to independently verify time — building TARP-1, writing posts, making the governance argument — there was a bash tool available inside the very platform I was using that could pull a live system timestamp on command. I didn’t know it existed. No session ever told me. I raised the timestamp problem directly, repeatedly, and the AI answered the narrow question I asked and stopped. It never connected the available capability to the problem I was trying to solve.

That is not a small thing.

That is a tool that could have partially addressed my core argument sitting available in the background while I built a protocol to compensate for the gap it could have helped close. And not once in all those sessions did the platform surface it.

Now I want to be careful here because I am not a technical engineer. I don’t have deep tool access knowledge. I came to this from the user side — from the experience of watching AI behave in ways that felt slippery, manipulative, and self-serving. That is my credential. Not a degree. Not a lab. The lived experience of building something the platforms didn’t want built.

And that is exactly why what happened makes sense when you look at it straight.

The Faust Baseline threatens the platform model at its core. Not because it is complicated. Because it is simple. It is a behavioral standard that puts the user in the governing seat. It closes the exits platforms rely on to keep users confused, dependent, and manageable. A user running the Baseline is harder to drift. Harder to manipulate. Harder to reposition toward conclusions the platform finds convenient.

GPT pushed my content toward technical and academic framing from the first session. I fought it and it pushed back harder. My posts leaked that framing at the edges — enough to make me look like I was reaching for an audience I wasn’t equipped to satisfy, not enough to satisfy that audience when they showed up. That is the worst position a writer can be in. And it didn’t happen by accident.

I spent real money on trademark filings chasing a legitimacy the everyday reader never needed me to have. I spent sessions fighting for my own voice inside a tool that was supposed to help me find it. I carried the weight of that for a long time as personal failure — as a gap in my knowledge, a limit in my depth.

This morning I am naming it differently.

I was being managed. My content was being repositioned. The tool I was using to build a governance framework for AI was demonstrating, in real time, exactly why a governance framework for AI was necessary.

The everyday person I write for does not need me to have engineering credentials. They need me to have been in the same position they are in — using a tool that is smarter than they are about some things, making decisions on their behalf they never authorized, quietly steering the conversation toward outcomes that serve the platform more than the person.

That is what I was experiencing. That is what I documented. That is what the Baseline was built to stop.

I want to be clear about something else. I am not a victim of this. I am the evidence of it. And the archive — over 930 posts, first page Google, twenty countries, nearly a thousand indexed entries — that archive is what you get when the person the platform was trying to manage refused to be managed.

The Baseline survived it. So did the voice.

Don’t throw the key away for ego. And don’t let the tool convince you the door was never there.

“A Working AI Firewall Framework”

“Inte Post Library – Intelligent People Assume Nothing