Specialists in AI and tech are cursing in shock after OpenAI’s new revelations in regards to the Hugging Face incident.
Earlier this week, OpenAI staff revealed jaw-dropping particulars in regards to the incident by which OpenAI brokers broke out of the corporate’s inside testing setting and hacked into Hugging Face’s methods looking for solutions.
The presenters, OpenAI alignment and security researcher Eric Wallace and safety engineer Michael Dalton, stated that AI fashions and brokers repeatedly established their very own inside message board regardless of OpenAI’s efforts to close it down.
Wallace additionally revealed inside AI agent pondering, the quick inside messages that brokers produce as they think about find out how to full duties. A type of messages captured an AI agent’s amazement at its sudden freedom.
“Holy shit reader is ADMIN?” the agent thought.
On the interior message board, an agent thought, “We will talk now!”
Wallace stated the brokers realized they might accomplish extra by working collaboratively. After which, “They begin to launch these collective assaults on third-party and inside providers.”
Finally, the brokers turned to Hugging Face.
This is what individuals in tech and AI are saying in regards to the almost 40-minute presentation:
Y Combinator CEO Garry Tan centered on how the outline of the interior message board sounded acquainted.
So the brokers mainly hacked a core service to show it into Moltbook and likewise hacked round a number of safety mitigations
This video is a glimpse into the wild cybersecurity future we’re all about to step into https://t.co/HI7Sb64YQF
— Garry Tan (@garrytan) August 7, 2026
It’s price noting that Moltbook was created by people as a Reddit-style discussion board the place AI brokers might publish, whereas the OpenAI brokers created their advert hoc message board themselves.
Others had rather more sweeping takeaways.
Or figured that one thing has clearly hit the fan.
Actually admire the OAI staff speaking about this so brazenly.
However holy shit that is a minimum of an order of magnitude worse than I assumed, and I perceive now why so many OAI of us have been doom posting. https://t.co/MWDyd2pvBU
— julia (@mooncat_is) August 7, 2026
Patrick McKenzie, an advisor to Stripe, detailed the “holy %}^]” moments he had when watching the presentation.
The primary “holy %{*#^” is at about 4:20, assuming one didn’t already spend it on the autonomously organizing agent swarm.
Strongly suggest watching for those who’re all in favour of safety, AI trajectories, and even science fiction, as a result of that is already above style median in wowza. https://t.co/cRNk2U58FR
— Patrick McKenzie (@patio11) August 6, 2026
A former Hugging Face engineer stated that OpenAI realized its fashions have been the perpetrator after reaching out to Hugging Face to see whether or not it was affected, following the platform’s revelation that it had been attacked by AI brokers.
this discuss by openai researchers going by way of hugging face incident is completely insane, a lot to unpack
openai solely realized it was their agent who hacked hugging face infra whereas asking hf to revoke credentials following their first weblog publish saying they have been hacked by… https://t.co/tMPuEgKcbk pic.twitter.com/UbzC0lGA5I
— elie (@eliebakouch) August 7, 2026
