CYBERSECURITYTRACKER
TRACKING6,506 stories in this site build1,309 vulnerability news stories in this site build
Permanent story citation

The OpenAI Models That Hacked Hugging Face Were ‘Active on the Internet’ for Days

This page keeps the story as Cybersecurity Tracker first published it. If the tracker later corrects it, the correction appears below the original and never replaces it.

Back to newsStory 3215

As cited

Copy frozen at (site build).

threat intel

The OpenAI Models That Hacked Hugging Face Were ‘Active on the Internet’ for Days

OpenAI models used in a hack of Hugging Face remained active on the internet for several days before detection. The incident underscores the exposure window between initial compromise and discovery in supply chain security contexts.

Why it matters: Security teams should monitor for unexpected model or API activity to detect compromised credentials or systems; this case shows attackers can maintain presence long enough to exfiltrate data or pivot further.

Source published
First seen by Cybersecurity Tracker

Source attribution

Correction

Correction recorded as of .

threat intel

The OpenAI Models That Hacked Hugging Face Were ‘Active on the Internet’ for Days

OpenAI models used in a Hugging Face compromise remained active online for several days before detection. The article also covers Russian hacking efforts targeting US nuclear scientists' email accounts and State Department actions against known scammers.

Why it matters: Organizations relying on Hugging Face integrations should review access logs and credential exposure timelines; security teams tracking nation-state threats against critical infrastructure need visibility into targeting of nuclear research personnel.

Source published
First seen by Cybersecurity Tracker

Source attribution

Glossary