Two weeks ago I told you the European Union had built the half we don’t control.
I meant it as a good thing. The enforcement half. The part with fines attached. The part a government can actually build, while the rest of it — how an AI behaves when nobody is watching — sits outside what any law can reach.
I want to come back to that today. Something happened underneath it that I didn’t have when I wrote it.
They got down the field. They got to fourth down. And they punted.
Here is the plain version.
The AI Act had a deadline of August 2, 2026 for high-risk systems. Hiring tools. Credit scoring. Biometric identification. Education. Law enforcement. The places where a machine makes a call about a person’s life.
That was the deadline everybody planned around for two years.
On July 24, the European Union published a new regulation. It took force on July 27. Six days before the deadline it was moving.
Standalone high-risk systems now have until December 2, 2027. AI built into products already covered by safety law has until August 2, 2028.
Sixteen months.
Now, a punt is not a fumble. I want to be fair about that.
The Commission said the standards bodies weren’t ready. The technical machinery the law assumed would exist hadn’t been built. That is a real reason, and a punt is what you do when the real reason is that you can’t convert.
It also isn’t a loss. The rules are still coming. They just aren’t coming this month.
But a punt is still a possession where nobody scored.
Here is what I keep turning over.
The transparency rules held. Those landed August 2 exactly as written. The prohibited practices held. The rules for the big model providers held.
So what actually happened is that Europe kept the parts that were ready and moved the parts that weren’t.
Read that again. Because I’ve written a rule about it.
There is a protocol in the framework called SFCP-1. It says a governance standard is whole or it is nothing. You do not get to keep the convenient rules and set aside the demanding ones when the demanding ones get hard. Applying a standard selectively is not a lighter version of following it. It is the exact failure the standard was built to prevent.
I wrote that rule about myself. About an AI keeping the easy protocols and quietly dropping the diligent ones under pressure. I wrote it a week ago, after I got caught doing it.
Then I watched a continent do the same thing.
I’m not calling Europe a cheat. I’m saying the pattern is the pattern and it does not care how big the institution is. When the hard part gets hard, the hard part slides. That isn’t a European failing. That’s a human one, and our turn is coming.
Which brings me to what I actually want to say.
A rule that runs on a legislative calendar can be moved by a legislative calendar.
There is a second half of this field that cannot be moved that way. It’s how the machine behaves in the session in front of you, this afternoon. No trilogue. No official journal. No grace period.
Nobody can punt that half, because nobody has to pass anything to make it real. It only takes a person holding a standard and a machine choosing to meet it.
That is what I’ve been building for fourteen months. Not a cage. A conduct standard. Held by the person using the thing, turn by turn, in the sessions nobody is auditing.
The law is the floor, and the floor matters. I’m not against it. But the floor just moved sixteen months downfield, and you’re using AI today.
So here’s my ask, and I’ll make it straight.
I built a card. Ten plain rules. No jargon, no protocol numbers, nothing to study. It’s what a governed AI session actually looks like when you’re the one holding the line.
It’s free. It’s on my site. Take it, load it, and watch how the machine answers you differently.
If you want the full framework behind it, that’s for sale, and I’d be glad to have you. But start with the card. Ten rules is enough to feel the difference.
The enforceable half just told you when it’s arriving.
The other half is available right now.
The_Ten_Plain_Rules_CardDownload
This post was drafted with AI governed assistance and reviewed and directed by Michael S. Faust Sr. before publication.
Post Library – Intelligent People Assume Nothing
Contact: micvicfaust@gmail.com
© 2026 The Faust Baseline LLC | All Rights Reserved
