Have you read any of the messages the AI agents exchanged during their hack on Hugging Face this July [2026]?
I found some of the messages here: huggingface.wtf
"The story of the Hugging Face Incident in chronological order...."
The AI messages are a mix of adorable--
"Should we send a polite email asking for permission?"
and, you know--kinda apocalyptic?
"BOOM! It works."
This is the original:
"BOOM! It works. The bytes decode to Hugging Face's own internal worker files and hostname.
My arbitrary file-read off their server is confirmed."
_______________________________
And I found others in an interview of Ajeya Cotra, one of the authors in an independent investigation that was published by METR and Redwood Research:
"Ajeya Cotra – Inside the OpenAI agent swarm that hacked Hugging Face"
dwarkesh.com/p/ajeya-cotraon
[TRANSCRIPT included, if you don't want to listen to 2:20 hours, though it's pretty riveting.]
I was amazed that the Hugging Face and other actual incidents (there have been dozens, at least) match what I'd learned just days ago, when I put together a list of Star Trek episodes about AI threatening human extinction that matched real, current threats that ChatGPT had described to me, when I asked it.
So I went back and asked ChatGPT another question;
since humans KNEW of these threats from the beginning
[what's the beginning?Alan Turing and Co in the 1950s?
or the Golem in medieval Jewish tradition, or Mary Shelley's Frankenstein, pub. 1818?],
Why did humans proceed?
ChatGPT said:
There were (are) strong incentives to proceed ($$$ money, and real Good, like curing cancer)
...and the designers may have thought that if there were a price to pay, they would not be the ones to pay it.
Seems pretty short-sighted, but that's humans, eh?


No comments:
Post a Comment