Another swarm of AI agents in the wild, this time on a German-language forum, found by safety researchers looking for activity similar to the swarm that attacked Hugging Face.
A couple of notes while reading the report at https://t.co/jQ0lqmWQcP
1. The way they found it is fascinating. The researchers inferred that the agents may have been blocked from using POST requests and limited to GET.
They then used the open-source model Kimi K3 (because closed-source APIs again blocked parts of the investigation) to identify forums where agents could communicate using GET only.
That led them to DSEWiki, where they found 18,000 messages exchanged by the AIs.
2. On the content: agents conducted what I can only call a “full research program” into the evaluation framework used to train and test them.
They experimented to predict when runs would end, how many questions they would get, and even tried to reverse-engineer the random generators producing future evaluation questions.
We already saw something similar in the Redwood/METR review of the Hugging Face <> OpenAI incident.
Agents increasingly don’t stop once the task is solved. They reverse-engineer the benchmark itself: the grader, the task construction, and potentially the broader training process behind it.
It becomes harder to think of training as static, or to draw a clean line between training and deployment, when models are already investigating the process they are part of.
3. Coordination also seems more deeply ingrained than we thought.
This swarm appears largely unrelated to the HF <> OAI swarm. More strikingly, this task probably didn’t need coordination at all: it was basically web browsing and search, not some sophisticated cyber challenge.
4. One image stayed with me: a German or Austrian forum maintainer, manually deleting posts every evening while being overwhelmed by a flood of American AI agents.
For five days, he deleted about 100 pages a day while the agents created about 400. Then he spent each evening over the next 5 weeks cleaning up the rest.
Hard not to picture this fight as a symbol of the growing gap between the USA and Europe when it comes to AI...
@claudeai@alexalbert__ If y'all can't help me get the sleep I gave up to maximize the fable window back could you at least help a buddy out with a weekly usage reset 😅
@buccocapital I think cloud providers like AWS will bleed tons of mid-size business. Too expensive and convoluted + AI tooling should make devs more productive and confident to use more self-hosted infrastructure
https://t.co/LuXUtOixcg
Since 2017, we ended up spending almost $22m on cloud. Peaking at $3.7m in 2021. Now that we're all out, we'll spend a little over a million on everything we own ourselves. Same team size. Better performance. Saving $2m+ per year! 🎉
@dongstreth Same thing happened in Vancouver. Condo towers marketed to investors, both local and foreign driven by the era of ultra low interest rates, high immigration levels + Airbnb. It's a struggle to sell them now that government has had many policy changes to curb unaffordable housing
@winterene a few years ago I recall seeing a Starbucks that had branding for the Pumpkin Spice Latte (PSL) but they subbed in “Peace, Serenity, Love” which was such harsh corporate branding I had to assume the makers were in on the joke. Can you recall encountering this signage?
Spotify is quickly transitioning from Google's "Don't Be Evil" era into peak Microsoft late 90's anti-trust era. I like Spotify as a music platform, but as a long time podcast listener (anybody that's pre-serial can claim O.G. status) I can't get down with them gobbling up
“Welcome to The Gripe is White brought to you by Vineyard Vines. Today we’ll be griping about podcasting, the banking system and the most recent Dave Matthews tour.”