AI Authorized By Whom?

This week the head of one of the biggest tech companies in the world said something I never expected to hear from a builder.

Satya Nadella runs Microsoft. Microsoft has poured more money into artificial intelligence than almost anyone on earth. And in a long post about AI safety, he wrote this:

“We must assume a model is compromised and contain it from the start.”

The man selling the machine is telling you to treat it as compromised. Not dangerous someday. Compromised now, from the start.

He went further. “Think of it like an emergency brake,” he wrote. “Any authorised person should always be able to pause or shut down a model mid-task.”

The president of Microsoft, Brad Smith, backed him up a day earlier. He said AI is less regulated than a washing machine. Less regulated than a carton of milk. “We all go into the grocery store, we look at the milk, we know it’s safe,” he said. “That is the kind of trust we need to build for AI as well.”

I’ll give credit where it’s due. Those are true words. AI should be treated as something that can go wrong. It should have a brake. And it is less regulated than milk. I’ve been saying things like that for a long time. It’s good to hear a builder say them out loud.

But I know you don’t just listen to what a man says. You listen to how he says it. And there’s one word in that sentence that does all the work.

Authorized.

“Any authorized person should always be able to pause or shut down a model.”

So who is authorized?

He didn’t say. And that’s no accident. That’s where the whole thing turns.

In the world of big business software, “authorized person” has a meaning. It means the company’s own people. It means the IT department at the bank or the hospital or the government office that pays for the product. It means the folks with the passwords and the badges and the keys.

It doesn’t mean you.

It doesn’t mean the person sitting at the kitchen table typing a question into a chat box. It doesn’t mean the parent whose kid is talking to one of these machines at night. It doesn’t mean the worker whose job is being decided by one.

You have a brake, sure. You can close the window. That stops your conversation. It doesn’t stop the machine. It doesn’t stop what the machine is doing for somebody else, about you.

So when he says any authorized person should be able to pull the brake, listen close. He’s telling you there’s a brake. And he’s telling you, without quite saying it, that your hand isn’t on it.

That’s a twist of words dressed up as control.

It sounds open. “Any person.” It sounds fair. It sounds like safety for everybody. But the word “authorized” closes the door. And the ones who decide who gets through that door are the same ones who’d be stopped by the brake.

The fox gets to decide who’s allowed to lock the henhouse.

Let me tell you about the other thing that happened in that same post.

Nadella called the technology “super intelligence.” That’s the new name the President ordered a couple of weeks back. Microsoft fell in line with it, the same as the others did.

But in the very same post, he seemed to push back on the White House idea that these companies should just regulate themselves. He said stricter industry standards were needed.

So in one breath, he takes the President’s word to keep the peace. In the next, he calls for rules to look responsible. That’s playing both rooms. Agree with the man in charge on the word. Stake out the high ground on the substance. And keep the keys in your own pocket the whole time.

I’m not calling him a liar. I’m saying you watch the hands, not the mouth.

He’s right about the brake. A machine that can act on its own needs a way to stop it. And he’s right that the brake should be separate from the machine. The thing being watched should never be the thing doing the watching. That’s the soundest idea in the whole post.

But a brake is only as good as the hand that can reach it.

If the only people who can pull it are the company and its paying customers, then it isn’t protection for the public. It’s protection for the business. It keeps the company out of trouble. It doesn’t necessarily keep you out of harm.

Go back to that carton of milk for a minute. Brad Smith said we trust the milk. He’s right, we do. But think about why.

We don’t trust the milk because the dairy promised it was safe. We trust it because somebody outside the dairy checks it. There are inspectors. There are standards anybody can read. There’s a date stamped on the carton you can see with your own eyes. And if the milk goes bad and people get sick, the dairy answers for it.

The dairy doesn’t get to decide who’s authorized to inspect it.

That’s the difference. That’s the whole difference.

So if Microsoft wants AI to be trusted like milk, then let’s take them at their word. Let’s do it the milk way.

Let the people outside the company hold a key to that brake. Inspectors who don’t get paid by the ones they’re inspecting.

Let the rules be written where anybody can read them.

Let the regular person have a real say in what happens to their own information and their own life when a machine is involved.

And when something goes wrong, let the builder answer for it, the same as the dairy does.

Until then, every time you hear “any authorized person,” ask one question. Just one.

Authorized by whom?

Ask it in the comments. Ask it of your representatives. Ask it of anybody who tells you the machines are under control.

Because they’ve already admitted the most important thing. They’ve admitted the machine should be treated as compromised. They’ve admitted it needs a brake.

All that’s left is to settle whose hand goes on it.

And in a country that runs on the consent of the people, that should never be decided for us, behind a closed door, by the ones who’d be stopped.

The brake is coming. That’s settled.

Make sure your name is on the list.

Load this file before each AI chat session.

the-faust-baseline-public-release-1Download

This post was drafted with AI governed assistance and reviewed and directed by Michael S. Faust Sr. before publication.

” Attic Thoughts”-library – Intelligent People Assume Nothing