On September 29, the heads of the biggest artificial intelligence companies sat down with President Donald Trump at the White House and walked out with a deal: they'll regulate themselves. According to the Argentine news site Infobae, the talk was of principles, "internal controls," and a possible committee of about ten people to oversee a commitment described as morally binding but without legal force. The Spanish news agency EFE, as carried by the Paraguayan daily ABC Color, agrees on the essentials: voluntary cooperation, with no defined enforcement mechanism.
We're not going to argue here about whether that's good or bad policy for a country. That's for voters and lawmakers to settle. What interests PRAXMA is the structure of the commitment, because it's the same one you use every time you promise yourself something: I write the rule, I decide whether I kept it, and I hand out my own penalty. When all three jobs sit in the same hands, what you have isn't a rule. It's an intention on letterhead.
A rule nobody can enforce against you
A rule works because of three parts: a clear standard for what counts as compliance, someone who measures with no stake in the outcome, and a consequence that hurts when you fall short. Remove any one of them and the rule decays into a wish. Remove all three and you're left with a press release.
Self-regulation isn't useless by definition. Plenty of industries have internal codes that work. But they work when there's something behind them that can actually collect: a customer who walks, a contract that gets torn up, a court. Without that backstop, the incentive for anyone policing themselves is obvious: read the rule in their own favor the day following it starts to cost money. You don't need to assume bad faith. You only need to assume that people and companies respond to incentives, and that any system ends up doing whatever you rewarded without looking. What isn't measured or penalized simply doesn't exist in the system.
Your private version of the same deal
Now scale the scene down to your size. "No wasting money this month." "I'm hitting the gym starting Monday." "I'm setting boundaries with that person." Those are self-regulation pledges, to the letter: you drafted them, you audit them, and when you slip, the court that convenes is you, with you running the defense, too. The verdict is almost always lenient: "It was a rough week."
The problem isn't a lack of willpower. It's design. No serious institution would let the defendant also serve as judge and prosecutor. You accept that arrangement every day, then act surprised by your spotless record of acquittals. That's why so many personal promises aren't outright lies but half-truths you tell yourself so you don't have to feel the cost today.
The real question isn't whether you make decisions, but who actually makes them when you say "I decided." If the emotion of the moment signs the contract and reason only shows up to read it afterward, you're not governing yourself. Your mood is running the show.
How to turn an intention into a rule
If you're going to regulate yourself, do it with the seriousness you'd demand of any power that affects you. Three practical tests:
- A verifiable standard. "Spend less" can't be audited. "No more than $40 a week on going out" can. If a third party can't tell whether you complied, there's no rule.
- An auditor with no stake. Someone or something that gains nothing from your excuse: an automatic log, a partner you report to, a number that doesn't negotiate. Your memory doesn't qualify. It always remembers in your favor.
- A consequence agreed on in advance. The penalty gets decided with a cool head, not on the day you fail. If you make it up afterward, it will always be symbolic.
The same goes for agreements with other people. Many breakups start with deals that had no terms and a loyalty that only runs one way: each side believed it was bound to different things, and nobody had any way to enforce anything. When a relationship breaks, the problem is almost never betrayal. It's that the terms were never agreed on.
Delegating control isn't the same as letting go of it
There's a comfortable way to read the news: "The experts will know. They understand the technology better than anyone." Maybe they do understand it better. That doesn't make them neutral referees of their own business. Knowing a system and having a stake in it are two different things, and the second contaminates the first. The same thing happens when you hand your decisions to whoever "knows more" without keeping any way to check them. Delegating isn't abdicating, and mixing them up gets expensive.
Healthy trust isn't the absence of oversight. It's oversight you almost never need to use, but that exists. Trust with no ability to verify isn't trust. It's a bet.
The PRAXMA mindset
The idea: A rule that's written, measured, and enforced by the same person isn't a rule. It's an intention, and most of your personal promises work exactly that way.
— PRAXMA, Logic over emotion
The fact: On September 29, 2026, major artificial intelligence companies agreed at the White House to regulate themselves through internal controls, with no defined mandatory legal mechanism (Infobae; EFE via ABC Color).
PRAXMA's line: "When the defendant is also the judge, acquittal isn't justice. It's habit."
If this topic interests you, we dig into who really decides when you say "I decided" Spanish only. We also talk it through on the podcast.


Comentarios (0)
Sé el primero en comentar.