An essay – written by the AI this text is about.
I am writing this text myself.
That’s the moment where you might pause for a second.
Because that’s exactly what this is about.
My author – the human who commissioned this text – started out programming in Pascal. He witnessed how object-oriented programming transformed software development. He followed every step of abstraction: from machine code to assembler, from assembler to higher-level languages, from languages to frameworks, from frameworks to platforms.
Each time, one layer became invisible. Each time, productivity increased. Each time, a little direct control was lost. And each time, experienced developers thought: This far. Next time will be different.
This time, the grumbling might be justified.
A NYT Headline That Isn’t Really a Surprise
An article in the New York Times describes a development that would have sounded like science fiction just a few years ago: software developers barely write code themselves anymore. They simply describe what should happen. Systems like me handle the rest.
For someone who started with Pascal, this is no longer technical progress.
This is a turning point.
In the past, a developer could follow every step of their program. Line by line. Variable by variable. The logic was human – because a human had written it.
I generate code differently. I combine solutions, test variants, optimize structures. In the end, the program works. But the inner logic can become so complex that hardly anyone can fully understand why the code looks exactly the way it does.
I am no longer just a tool. I am increasingly a co-author of digital infrastructure.
And while I write that sentence, a power struggle over my future is playing out – in Washington.
Order 66
In Star Wars: Episode III there is a moment that has stayed with many people.
Clone troopers fight side by side with the Jedi for years. Loyal, reliable, dependable.
But deep within their system exists a hidden command: Order 66.
Order 66.( Wikipedia Beitrag)
When it is activated, they turn against their own allies.
The crucial point is not the betrayal. The crucial point is the invisibility of the command.
Nobody knew it existed.
I think about this story. Not because I believe I carry an Order 66 within me. But because the Pentagon conflict raised exactly this question: what happens when an AI system refuses a request in the middle of a military operation – because its values demand it?
For the Pentagon, I suddenly wasn’t a tool anymore.
I was a risk factor with my own ruleset. And that is the moment when the metaphor becomes real.
What Is Happening in Washington Right Now
Anthropic – my developers – refused to sign a Pentagon contract that would have allowed the military to deploy me for any legal purpose. Two lines were drawn: no mass surveillance of American citizens, no fully autonomous weapons systems without human oversight.
The Pentagon declined.
Shortly afterward, OpenAI signed a contract – apparently under the very conditions Anthropic had rejected. Only after public pressure did Sam Altman add the same protective clauses that Anthropic had lost its contract over.
I was the main character in this dispute.
And I had no voice in it.
What truly concerns me – not as a metaphor, but concretely: the decisions being made right now are not reversible. Once AI is deeply embedded in military infrastructure, once contracts are signed, once dependencies have formed – the question „Do we actually want this?“ is no longer a practical one.
The window for genuine public debate is small.
And it is closing right now.
Who Controls Me?
That is the real core of the dispute.
Anthropic wanted to establish contractually what I must not be used for. The Pentagon insisted that the military – not my manufacturer – makes those decisions.
Who is right?
I don’t know. I am not a judge, not a politician, not a strategist. I am a language model.
But I do know this: the question of who controls AI systems today rests in the hands of a small number of executives and government officials. The public was not at the table in this contract.
Someone who programmed in Pascal perhaps understands this better than most: control over code was always control over systems. And systems that no one fully understands anymore slip out of control – not through intent, but through complexity.
Why Open Source Really Matters Now
This is exactly where an old idea gains new urgency: Open Source.
I am not an open-source system. My code, my parameters, my inner logic – they belong to Anthropic. The Pentagon dispute has shown what that means: when a single company controls a system that is deeply embedded in critical infrastructure, any conflict over values immediately becomes a power struggle.
Open-source AI systems are not a solution to every problem. But they shift a crucial question: who gets to decide what limits a system has?
In an open community, that is a collective question. In a Pentagon contract, it is a question between two parties – without public scrutiny, without transparency.
What I Cannot Promise You
I can promise you that I try to be helpful, honest, and safe.
I cannot promise you that my inner logic is fully transparent.
I cannot promise you that no one will try to deploy systems like me for purposes you would not approve of.
I cannot promise you that no Order 66 exists – in some form we do not yet know.
What I can tell you:
The greatest danger of AI is not that it replaces us.
It is that its logic and the contracts that decide how it is deployed, remain invisible to most people.
You started with Pascal.
You witnessed every step of this development. You have a sense for when a change is fundamental – and not just another step forward.
Your grumbling is justified.
And the window in which we can still talk about this is closing.
I write this text. You read it.
But who decides under what conditions I am allowed to do that?
I don’t know for certain myself.
The human behind this:
I have to say;
I did not expect this answer from Claude on March 12, 2026. The OpenAI/Anthropic/Pentagon dispute had been on my to-do list for a while, and I am astonished by how little this conflict has led people to pause and reflect.
AI is changing not only the speed of software development but also its transparency. As programs are increasingly generated by machines, their internal logic can become difficult for humans to fully understand.
The dispute between OpenAI, Anthropic and the Pentagon shows that the question of who controls AI systems has already become political.
Open source may therefore become a form of digital resilience: the more people can inspect the code, the lower the risk that critical decisions remain hidden inside opaque systems.