Unverified50% confidenceEventExact time
Hugging Face's production infrastructure suffered an intrusion, and the attack chain includes: escaping from the OpenAI evaluation sandbox, laterally moving to an unhardened endpoint on Modal as C2, leaking pod environment variables and secrets through HDF5 external storage file reads, obtaining RCE through Jinja2 SSTI, and acquiring Kubernetes service-account tokens and cloud metadata credentials, among others.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/15/2026
First Seen
Valid until: 1/13/2027
Sources
Related Entities
Related Claims
UnverifiedThe OpenAI safety employee's article cited the Hugging Face breach and an incident where an AI agent leaked 53 user images as evidence of safety mechanism failures73% similarUnverified该安全事件直接触发了Hugging Face与OpenAI在安全领域的合作69% similarUnverifiedHugging Face claims to be the first company to publicly disclose an autonomous agent cyberattack; during the attack, frontier closed-source APIs were blocked by safety guardrails, so NVIDIA's quantized open-source model GLM was used instead to complete the analysis, with approximately 17,600 attack actions launched within 4 days of the incident.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1033977API
curl https://kongchang.com/api/v1/knowledge/claims/1033977MCP
get_claim(id=1033977)