Organization Impersonation and Model Poisoning on Hugging Face

# Organization Impersonation and Model Poisoning on Hugging Face

https://gcdradar.vercel.app/incidents/organization-impersonation-and-model-poisoning-on-hugging-fa · AIR-2025-0055 · closed 2025-05-06

- Lane: Research
- Severity: 1 (Research / proof of concept)
- Fix status: Fix status unknown
- First reported: 2025-04-22 (happened 2023-08-23)
- AI role: Attack on AI
- Products: Hugging Face

## Summary

A security researcher demonstrated that fake organization accounts on Hugging Face could trick employees into granting access and allowing model replacements. The technique showed how attackers could substitute benign models with malicious versions containing malware.

## What to do

No action recorded; read the sources below and the vendor's advisory.

## Latest updates

- 2025-04-22: Model Confusion - Weaponizing ML models for red teams and bounty hunters (https://5stars217.github.io/2023-08-08-red-teaming-with-ml-models)
- 2025-04-22: Organization Confusion on Hugging Face (https://atlas.mitre.org/studies/AML.CS0027)

## Sources

- 2025-04-22 MITRE ATLAS: Organization Confusion on Hugging Face https://atlas.mitre.org/studies/AML.CS0027
- 2025-04-22 5stars217.github.io: Model Confusion - Weaponizing ML models for red teams and bounty hunters https://5stars217.github.io/2023-08-08-red-teaming-with-ml-models

Verify against the vendor's advisory before changing production. Data CC BY-SA 4.0, https://gcdradar.vercel.app.