Bug 2376775 (CVE-2025-3777)

Summary:	CVE-2025-3777 transformers: Improper Input Validation in huggingface/transformers
Product:	[Other] Security Response	Reporter:	OSIDB Bzimport <bzimport>
Component:	vulnerability	Assignee:	Product Security DevOps Team <prodsec-dev>
Status:	NEW ---	QA Contact:
Severity:	low	Docs Contact:
Priority:	low
Version:	unspecified	CC:	alinfoot, anpicker, bbrownin, bparees, dtrifiro, haoli, hasun, hkataria, jajackso, jcammara, jfula, jkoehler, jmitchel, jneedle, jowilson, jwong, kegrant, koliveir, kshier, lphiri, mabashia, nyancey, ometelka, pbraun, ptisnovs, rbryant, shvarugh, simaishi, smcdonal, stcannon, syedriko, teagle, tfister, thavo, ttakamiy, weaton, xdharmai, yguenane
Target Milestone:	---	Keywords:	Security
Target Release:	---
Hardware:	All
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	---
Doc Text:	An input validation flaw has been discovered in the Hugging Face Transformers library. Given an attacker can feed in a url, an application using the transformers library may present what appears to be a YouTube link, but actually routes to a malicious domain.	Story Points:	---
Clone Of:		Environment:
Last Closed:		Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description OSIDB Bzimport 2025-07-07 10:02:01 UTC

Hugging Face Transformers versions up to 4.49.0 are affected by an improper input validation vulnerability in the `image_utils.py` file. The vulnerability arises from insecure URL validation using the `startswith()` method, which can be bypassed through URL username injection. This allows attackers to craft URLs that appear to be from YouTube but resolve to malicious domains, potentially leading to phishing attacks, malware distribution, or data exfiltration. The issue is fixed in version 4.52.1.