CVE-2026-44513

Diffusers: `trust_remote_code` bypass via `custom_pipeline` and local custom components

HIGH
8.8
CVSS Score
Published: May 14, 2026 at 04:26 PM UTC
Updated: October 26, 2026 at 02:10 PM UTC
4 products affected across 2 vendorsDiffusersOpenShift AIEnterprise Linux AI

Using diffusers by huggingface?

Get instant alerts when new vulnerabilities affect diffusers - before attackers find them.

Real-time alertsTrack your software stackFree plan available
Severity
HIGH8.8
Exploit prediction
EPSS 1%
SSVC
No known exploit

Description

Diffusers is the a library for pretrained diffusion models. Prior to 0.38.0, a trust_remote_code bypass in DiffusionPipeline.from_pretrained allows arbitrary remote code execution despite the user passing trust_remote_code=False (or omitting it, which is the default). The vulnerability has three variants, all sharing the same root cause — the trust_remote_code gate was implemented inside DiffusionPipeline.download() rather than at the actual dynamic-module load site, so any code path that bypassed or short-circuited download() also bypassed the security check. DiffusionPipeline.from_pretrained('repoA', custom_pipeline='attacker/repoB', trust_remote_code=False) — the gate evaluated against repoA's file list rather than repoB's, so repoB's pipeline.py was loaded and executed. DiffusionPipeline.from_pretrained('/local/snapshot', custom_pipeline='attacker/repoB', trust_remote_code=False) — the local-path branch never invoked download(), so the gate was never reached and remote code from repoB executed. DiffusionPipeline.from_pretrained('/local/snapshot', trust_remote_code=False) where the snapshot contains custom component files (e.g. unet/my_unet_model.py) referenced from model_index.json — same root cause; the local path skipped download() and custom component code executed. This vulnerability is fixed in 0.38.0.

Affected

Our own answer: ranges merged, affected separated from platform-only, with confidence and review flags.

v < 0.38.0
Affected range stops below 0.38.0, so it looks safe — our inference, not a vendor statement
All versions
All versions
Vendor states this is fixed in 1786611435 (unaffected)
Vendor states this is fixed in 1786611803 (unaffected)
Vendor states this is fixed in 1787076481 (unaffected)
Vendor states this is fixed in 1787077779 (unaffected)

Remediation

No guide yet

No remediation guidance has been written for this CVE yet. What the record does show:

References (5)

x_refsource_CONFIRMExploitMitigationVendor Advisory
Source: cverecord
https://github.com/huggingface/diffusers/security/advisories/GHSA-98h9-4798-4q5v
Source: 0b0ca135-0b70-47e7-9f44-1890c2a1c46c
https://access.redhat.com/security/cve/CVE-2026-44513
Source: 0b0ca135-0b70-47e7-9f44-1890c2a1c46c
https://bugzilla.redhat.com/show_bug.cgi?id=2477507
Source: 0b0ca135-0b70-47e7-9f44-1890c2a1c46c
https://access.redhat.com/errata/RHSA-2026:60520

SSVC — Action Prioritization

SSVCNo known exploitNot automatableTotal impact

SSVC signals action urgency (what to patch first), separate from CVSS which measures severity.

CVSS

8.8HIGHGitHub_M
Attack Vector
NETWORK
Complexity
LOW
Privileges
NONE
User Interaction
REQUIRED
Scope
UNCHANGED
Confidentiality
HIGH
Integrity
HIGH
Availability
HIGH
Vector?

Quick Info

StatusMODIFIED
CWE
CWE-94CWE-358

Get alerted

Stop reading CVE feeds looking for your own stack

Track the products you run and VulnTracker tells you when a new CVE affects them — with what is known about fixing it, and nothing about the rest.

Free plan, no card