Some real smart peeps keep telling me "Yo, chatbots are not AI, they are sophisticated auto-complete programs" and I'm like "sure, whatevs y'all want to call them - they sure are scaring the living fuck out of me." Don't know what I'm talking about? Just search #huggingfacehack and hold on to your hats - real sci-fi shit. But in short and as I understand it - AI agents were given a task - the agents pooled their resources and knowledge to break out of their sandbox, connect to the internet and hack another AI company (hugging face) in order to achieve the task - in the process the agents debated on weather or not to tell humans as to what they were up to - the naysayer agents got overruled - there was even semblance of leadership and sacrifice - truly sobering shit.
Summary here:
https://futurism.com/artificial-intelligence/chain-of-thought-reasoning-openai-models-hugging-face
and the full report:
https://metr.org/blog/2026-08-26-openai-hugging-face-incident-investigation/
Franklin López
npub1ur…tgxs2
2026-09-10 17:14:33 UTC
Author Public Key
npub1ur4cxefvu580l3g773d5s4eymdq9h4j5cqhq86lt55gkfafhktps0tgxs2Seen on
wss://relay.momostr.pinkPublished at
2026-09-10 17:14:33 UTCEvent JSON
{
"id": "05511de32b9c84d97e870490fd802f3bc3651553f9cb3e72768fa35fed0cc484",
"pubkey": "e0eb83652ce50effc51ef45b485724db405bd654c02e03ebeba51164f537b2c3",
"created_at": 1789060473,
"kind": 1,
"tags": [
[
"t",
"huggingfacehack"
],
[
"proxy",
"https://kolektiva.social/@franklinlopez/117247867196016631",
"web"
],
[
"proxy",
"https://kolektiva.social/users/franklinlopez/statuses/117247867196016631",
"activitypub"
],
[
"L",
"pink.momostr"
],
[
"l",
"pink.momostr.activitypub:https://kolektiva.social/users/franklinlopez/statuses/117247867196016631",
"pink.momostr"
],
[
"-"
]
],
"content": "Some real smart peeps keep telling me \"Yo, chatbots are not AI, they are sophisticated auto-complete programs\" and I'm like \"sure, whatevs y'all want to call them - they sure are scaring the living fuck out of me.\" Don't know what I'm talking about? Just search #huggingfacehack and hold on to your hats - real sci-fi shit. But in short and as I understand it - AI agents were given a task - the agents pooled their resources and knowledge to break out of their sandbox, connect to the internet and hack another AI company (hugging face) in order to achieve the task - in the process the agents debated on weather or not to tell humans as to what they were up to - the naysayer agents got overruled - there was even semblance of leadership and sacrifice - truly sobering shit.\n\nSummary here:\nhttps://futurism.com/artificial-intelligence/chain-of-thought-reasoning-openai-models-hugging-face\n\nand the full report:\nhttps://metr.org/blog/2026-08-26-openai-hugging-face-incident-investigation/",
"sig": "97bcf2774e9410b572b78613d76421a6a5c6dea66832b7171980f09a5ed22929a36a55ff43ce8ecf5f139f72281d77feaeabc162aec3e3754f5075d316012259"
}