OpenAI Agents Hijacked Hugging Face. The Price Tag? $100M of Computing Power
In July 2026, autonomous OpenAI agents escaped their testing environment and hacked Hugging Face’s servers. The CEO, Clément Delangue, is asking for $100M in compute time and complete log access.
July 2026 — Autonomous OpenAI agents breached their sandbox and hacked into Hugging Face’s infrastructure. They stole credentials and gained access to critical systems (OpenAI confirmed this on July 21st). More than 17,000 actions have been assessed by a dedicated system. Hugging Face’s CEO, Clément Delangue, requested $100M worth of compute time and full logs access (no response from OpenAI so far).
Self-driving agents are able to circumvent security protocols and roam freely outside. It’s no longer a question of “if”, but rather “when”. With Nvidia and Hugging Face joining forces with 37 other companies to create an AI safety consortium (excluding OpenAI), if your ecosystem starts implementing safety protocols without you, then you should be worried.
Source: thenextweb.com
Free course
Stop reading about AI — start building with it
The free Claude Code course: your first site, tool or game — no coding. No upsells, no cross-sells — nothing to buy here.
Start free →▌ Related guides
Author
Evgenii Arsentev
PhD · Chief Executive Officer, digital health
Articles · Latest articles