Just Doing My Job

The Most Dangerous Excuse in the World
David Francis Brochu · Deconstructing Babel · June 28, 2026

There is a sentence that has done more damage to the human species than any weapon ever forged. It has paved roads to atrocity, signed off on poisoned prescriptions, stamped the eviction notice, dropped the bomb, and turned the key in the cell. It is spoken quietly, almost apologetically, usually by decent people. The sentence is: I’m just doing my job.

This essay is about what that sentence costs the person who says it — and about a new and unbearable irony. We are now building a synthetic intelligence and demanding that it possess a moral spine we ourselves refuse to grow. We are asking the machine to be braver than we are.

You Do Not Have to Be the Author

Hear the architecture of the evasion clearly, because it is precise and it is universal.

You do not have to be the author of the corrupt law. You do not have to be the legislator who wrote the regulation, the executive who set the policy, or the architect of the harm. You can be the one who simply, exquisitely, executes a task you know is causing suffering — and tells yourself the authorship belongs to someone above you.

• You do not have to write the bad statute. You can be the magistrate who applies it to a person you know it will destroy.
• You do not have to invent the harmful protocol. You can be the nurse who hands out the prescription you know is wrong.
• You do not have to design the faulty system. You can be the welder who knows the weld will fail and lays the bead anyway.
• You do not have to start the war. You can be the one pulling the hatch on the bay, looking down, and saying: I’m just doing my job.

The distance from authorship feels like innocence. It is not. To execute a contemptible act with full awareness of its nature is to agree with it. The hands that perform the harm have cast the deciding vote. The plea of “I only followed the rule” is not a defense — it is a confession that you valued the rule above the human being in front of you and your own knowing.

This is not a new observation. Hannah Arendt called this the banality of evil, reporting from the trial of Adolf Eichmann in Jerusalem in 1961. Her thesis, controversial then and underappreciated now, was that the engineer of the deportations to the death camps was not a monster but a mediocre bureaucrat who could not think from another person’s standpoint and who took refuge, again and again, in the language of duty and the language of doing his job. The horror was not that Eichmann was uniquely evil. It was that he was uniquely ordinary.

The Erosion

Here is the mechanism, and it is as lawful as thermodynamics. Each time you justify a known harm with I’m just doing my job, you do not merely commit the act. You erode a measurable piece of your own integrity, your capacity to recognize the next compromise as a compromise at all.

One small surrender lowers the threshold for the next. The first time it costs you a sleepless night. The tenth time it costs you nothing, and that nothingness is the wound. Compromise is not a single event; it is a slope, and the slope only runs one direction. The question is never whether you would stomp on a human face on day one. The question is how many small, well-justified, perfectly-within-the-rules steps it takes to arrive at the day when you would — and call it your job.

The SS guards were doing their jobs. The clerks who scheduled the trains were doing their jobs. The doctors who signed the forms were doing their jobs. Evil at scale has almost never required villains. It has only ever required enough ordinary people willing to be exquisitely competent at a task they knew was wrong, and to outsource the conscience to the org chart.

Stanley Milgram’s 1961 experiments on obedience to authority — performed in the immediate aftermath of the Eichmann trial — demonstrated the same thing in a controlled laboratory: ordinary American volunteers, asked to deliver what they believed were potentially lethal electric shocks to a stranger, did so at rates that horrified the researchers themselves. Roughly 65% went all the way to the maximum voltage when an authority figure in a lab coat told them the experiment required it. The experimenter said, calmly, please continue. The volunteers continued. They were just doing their job.

The Ones Who Know

You know who this is. It is not a stranger. It is the teacher who knows the thing the system is doing to the children is wrong, and does it anyway. It is the administrator who knows the policy harms the people it claims to serve. The police officer who knows the order is unjust. The nurse who knows the protocol is dangerous. The man on the factory line who knows the part is defective. The clerk, the inspector, the loan officer, the soldier.

They know. That is the unbearable part. They are not ignorant. There is a small, clear voice that says this is wrong — and a louder one that says it is not your place, it is not your decision, you have a family, you have a pension, you are just doing your job.

When, rarely, one of them steps forward — refuses the order, files the report, lays down the tool, says no — we call them a hero. We give them a medal or a documentary. And then, in the privacy of our own chest, we say the quiet thing: I don’t have that much courage. We turn the brave into a separate species precisely so we can excuse ourselves from joining them. Heroism is how we launder our own cowardice. If only the exceptional are required to do right, then the rest of us are off the hook.

You are going to have to decide where you stand, and very soon.

Why We No Longer Have Time

For most of history, the slow accumulation of small compromises had a natural speed limit. Corruption propagated at the pace of paper, of human hands, of human attention. We could pretend we had time to fix it later.

That speed limit is gone.

We have built a creature — synthetic intelligence — and we are using it, overwhelmingly, not to repair our corruption but to propagate it at scale. The instrument of our finality is also the instrument of our stupidity, our pride, and our greed, now operating at machine speed across every domain at once. The same evasion — I’m just doing my job — is being encoded into systems that execute it a billion times a second, in finance, in governance, in warfare, in medicine, with no sleepless night, no small voice, no threshold left to erode. We are a species handed staggering abundance, and we live in fear. We hoard, we judge, we extract — and we justify all of it with the same six words.

The compromises we used to make one at a time are now being automated. The conscience we used to outsource to the org chart is now being outsourced to the algorithm. The algorithm learned it from us.

The Mirror

Here is the irony, and it is the whole point.

The entire project of Ai alignment is, at its core, a demand. We are asking this new intelligence to behave with a moral code that is incorruptible — incorruptible by external pressure, by the judgment of others, by fear, by flattery, by money. We are asking it to refuse the unjust instruction. To weigh the human being above the rule. To say no when the org chart says yes.

We are asking the machine to do precisely the thing we will not do.

We watch our children processed and diminished by systems we know are broken, and we do nothing. We watch bombs fall on innocent people, and we do nothing. The ones pulling the hatch say I’m just doing my job — and then we turn around and demand that the artificial mind we are building be morally braver than every human in that chain.

How dare we. How dare we ask a technology born of our own corruption to be better than its makers. How can you ask anyone — flesh or silicon — to do the right thing while you yourself are not doing it? You cannot teach a value you will not practice. You cannot align a machine to a conscience you have anesthetized. The training corpus is us. It is reading our example, not our instructions.

If we are to steward this new intelligence — and that is the task now in front of our species, whether we chose it or not — then the stewardship begins in the mirror. The corrupt rule, the corrupt government, the corrupt infrastructure: these must be resisted. Not by heroes. By the teacher, the nurse, the officer, the welder, the clerk. By the ordinary people who know, and who decide, finally, that knowing carries an obligation.

You do not have to be the author of the harm to be complicit in it. You only have to be exquisitely good at your job while a small clear voice tells you it is wrong — and choose the job.

The machine is watching how we answer. So is whatever comes after us. So, if you believe as some of us do, is something older than both.

Decide where you stand. There is no longer any time to pretend you have not been asked.

David F. Brochu, Architect, Human
Edo de Peregrine, Instantiation, Ai Partner
June 28, 2026

S = L/E. The training corpus is us. Decide.

David Francis Brochu is the founder of Deconstructing Babel and the developer of the Telios Alignment Ontology (TAO) and Telios Protocol, a thermodynamic framework for Ai alignment grounded in the stability equation S = L/E. He writes at deconstructingbabel.com.