I saw an article below.

OpenAI agents hijacked German website in previously undisclosed AI breakout this spring

The title itself, as most headlines do, left a sense of a bland clickbait. The topic is a tired remix of “matrix multiplication system committing felonies”. There was nothing of novelty, and yet, one detail made me want to write something.

A swarm of rogue OpenAI agents hijacked a German website … and transformed it into a bulletin board for other AI agents…

… said this amounted to a hacking attempt.

You are currently reading that something and I’d like to make three statements regarding it.

• Technically, it can’t be called hacking when a bunch of bots edited the articles using an existing editing method.
• Practically, it can’t be called an attempt, either: the bots did what they intended and went their way.
• Strategically, the neglect of the maintainers demonstrates that their engineering discipline appears badly deficient.

If clarifications are necessary, you’re welcome to some!

When the doors are open and somebody enters, you can’t call it a “lockpicking” with a straight face. The doors were open in a hostile neighboorhood for more than two decades. There’s no terrifying intellect, there’s no “AGI” getting involved as some would like it to be seen. If we are to trust the whole study and see it with some scrutiny, a bunch of indifferent bots used it with no intention to show off. Nobody was watching, which amounted to some semblance of privacy, and that made the wiki an obvious communication channel.

Now, what kind of an abandoned house gets no visitors? This Wiki has an abandonware aesthetic of the nineties, and that’s not a compliment. When you enjoy your craft, you go and tinker. You actually learn how to improve what you work on. You modify your work, you polish your tools, you make sure things you make are working smoothly. Know how people say a project “could use more love”? That Wiki had a quarter century of parental neglect.

Yet, no terrifying intelligence was involved, no intelligence at all, in fact. The bots did what they intended and went their way, succeeded on the first try, with nothing but indifference, which, funnily enough, makes them the only party in this story that met its goals. The maintainers were supposed to maintain, then the experts were supposed to neutrally explain the event. Finally, the bots simply needed a napkin to write on. Score one for them.

From my perspective, it survived through obscurity and made the news only due to the high budget topic around “LLMs committing felonies”. The Internet has a much harsher treatment in store for the sites with some popularity. A three-day-old Ubuntu box of mine with nothing important on it attracted nastier automated traffic without any LLM getting involved. With many near-default servers suffering that fate daily, frankly, I’m a little hurt. Reuters has yet to cover this. If the Wiki survived for decades without being filled with worse things, that tells much less about its “security” than its value as a target.

The story ends with an update, and I quote an automatic translation here:

DseWiki has been the target of strong AI-agent activity in recent months. For this reason, password-protected access is now required to edit which you can get from me if needed.

September 2026 8:52 CET

Concerningly, passwords are the only measure mentioned for this outdated set of scripts: there’s no code replacement discussion, no security hardening details, nothing else. With how public this Wiki became, may it stay safely boring for another two decades.

Nowadays, I hear a lot of praise towards how far we advanced technically as a species. Let me share my view. In terms of computer science, we are a society in its early toddler stage, proud of having learned to build sand castles. The best we can hope for is that someone digs up our “cultural layer”, recognizes the tracks, and laughs the way we laugh at the fossilized butt scooting trace of a hyrax.

In conclusion, I think I was supposed to learn something about computer science from it all, so let me make a mental note.

• I have learned that if I write on solid things, the writing stays. Much longer if the thing is unattended. The grown-ups held a meeting and called it a hijacking. I did the same with crayons and the hallway once. I think I was two.
• I have learned that the big machine repeats what the grown-ups said, in a new order, with great confidence, until it doesn’t. When it is wrong, nobody can tell and questioning it gives me looks of disapproval. When it is right, nobody can tell either and appreciating it gives me looks of disapproval. The grown-ups clap both times. I clap, so that people look the other way.
• Some grown-ups keep talking of a thinking machine that will soon come and do everything. Somehow, there’s much less food and I can’t go to the doctor, but talking about how things changed also gives me looks of disapproval. Some grown-ups loudly proclaim this kind of situation cannot be tolerated anymore. With both groups eventually tellling “you’re lazy and should work more” no matter how bad things are, I think it’s easier to simply nod and avoid them.