CVE-2026-68770

A flaw was found in sentence-transformers. A remote attacker could exploit a logic flaw to bypass a security control designed to prevent the execution of untrusted code. This vulnerability allows for arbitrary code execution when an application loads a model from a directory that has been...

Hugging Face Diffusers Flaws Could Let Model Repositories Execute Arbitrary Code

Three high-severity security flaws have been disclosed in Hugging Face's Diffusers library that could allow crafted model repositories to stealthily execute arbitrary code on machines that load it, opening the artificial intelligence AI supply chain to security risk. "These vulnerabilities are...

CVE-2026-68770

sentence-transformers contains a security control bypass vulnerability that allows attackers to achieve arbitrary code execution by exploiting a logic flaw in the importmoduleclass helper within sentencetransformers/util/misc.py, where the guard condition includes an 'or...

CVE-2026-68770

sentence-transformers contains a security control bypass vulnerability that allows attackers to achieve arbitrary code execution by exploiting a logic flaw in the importmoduleclass helper within sentencetransformers/util/misc.py, where the guard condition includes an 'or...

CVE-2026-68770 sentence-transformers Arbitrary Code Execution on Local Model Load Despite trust_remote_code=False

sentence-transformers contains a security control bypass vulnerability that allows attackers to achieve arbitrary code execution by exploiting a logic flaw in the importmoduleclass helper within sentencetransformers/util/misc.py, where the guard condition includes an 'or...

CVE-2026-68770 sentence-transformers Arbitrary Code Execution on Local Model Load Despite trust_remote_code=False

sentence-transformers contains a security control bypass vulnerability that allows attackers to achieve arbitrary code execution by exploiting a logic flaw in the importmoduleclass helper within sentencetransformers/util/misc.py, where the guard condition includes an 'or...

EUVD-2026-51650

sentence-transformers contains a security control bypass vulnerability that allows attackers to achieve arbitrary code execution by exploiting a logic flaw in the importmoduleclass helper within sentencetransformers/util/misc.py, where the guard condition includes an 'or...

CVE-2026-68770

The CVE affects the sentence-transformers library, where a logic flaw in the import_module_class helper (sentence_transformers/util/misc.py) adds an incorrect guard: os.path.exists(model_name_or_path) causes the trust_remote_code gate to pass whenever the provided path exists locally. This enable...

PT-2026-66862

Name of the Vulnerable Software and Affected Versions sentence-transformers affected versions not specified Description A security control bypass exists due to a logic flaw in the import module class helper within sentence transformers/util/misc.py. The guard condition includes an or...

CVE-2026-45804

Diffusers is the a library for pretrained diffusion models. Prior to 0.38.0, Diffusers' DiffusionPipeline.frompretrained flow can bypass the trustremotecode guard because download validates modelindex.json and custom pipeline code before later loading from a cached folder that can change, allowin...

CVE-2026-45804 Diffusers: TOCTOU Trust Remote Code Bypass

Diffusers is the a library for pretrained diffusion models. Prior to 0.38.0, Diffusers' DiffusionPipeline.frompretrained flow can bypass the trustremotecode guard because download validates modelindex.json and custom pipeline code before later loading from a cached folder that can change, allowin...

CVE-2026-45804 Diffusers: TOCTOU Trust Remote Code Bypass

Diffusers is the a library for pretrained diffusion models. Prior to 0.38.0, Diffusers' DiffusionPipeline.frompretrained flow can bypass the trustremotecode guard because download validates modelindex.json and custom pipeline code before later loading from a cached folder that can change, allowin...

CVE-2026-45804

Diffusers CVE-2026-45804 describes a TOCTOU race in DiffusionPipeline.from_pretrained where two Hub fetches (config via hf_hub_download and a snapshot via snapshot_download) can resolve to different commits. An attacker could push commit B after the first check, causing the newer commit’s model_i...

PYSEC-2026-2608 lmdeploy: Hardcoded trust_remote_code=True is an implicit unsafe remote-code load path with no user opt-out

📋 Reframing 2026-05-02: implicit unsafe remote-code path, not "supply-chain" The accurate description of this vulnerability is: "getmodelarch and related helpers hardcode trustremotecode=True with no opt-out, creating an implicit unsafe remote-code load path on every model fetch." What this repor...

PYSEC-2026-2609 LMDeploy: Arbitrary code execution via hardcoded trust_remote_code=True in lmdeploy model initialization

Summary lmdeploy hardcodes trustremotecode=True in multiple HuggingFace model-loading call sites. The affected code paths are in: text lmdeploy/archs.py lmdeploy/utils.py The vulnerable call sites pass trustremotecode=True into HuggingFace Transformers APIs such as AutoConfig.frompretrained,...

PYSEC-2026-2446 Diffusers: TOCTOU Trust Remote Code Bypass

Background This vulnerability is found in the diffusers package - the transformers-equivalent library for diffusion models. It is found in the DiffusionPipeline.frompretrained flow, which is used to load a pipeline from the HuggingFace Hub. This function has a trustremotecode guard: if the...

PYSEC-2026-2522 InstructLab Includes Functionality from Untrusted Control Sphere

A flaw was found in InstructLab. The linuxtrain.py script hardcodes trustremotecode=True when loading models from HuggingFace. This allows a remote attacker to achieve arbitrary Python code execution by convincing a user to run ilab train/download/generate with a specially crafted malicious model...

PYSEC-2026-2010 vLLM affected by RCE via auto_map dynamic module loading during model initialization

Summary vLLM loads Hugging Face automap dynamic modules during model resolution without gating on trustremotecode, allowing attacker-controlled Python code in a model repo/path to execute at server startup. --- Impact An attacker who can influence the model repo/path local directory or remote...

EUVD-2026-40311

LLaMA-Factory through 0.9.5 contains a remote code execution vulnerability that allows attackers with WebUI access to execute arbitrary Python code by supplying a malicious model path in the Chat or Training interfaces. The application passes user-supplied model path input unvalidated into...

CVE-2026-58116 LLaMA-Factory 0.9.5 Remote Code Execution via WebUI Model Path

LLaMA-Factory through 0.9.5 contains a remote code execution vulnerability that allows attackers with WebUI access to execute arbitrary Python code by supplying a malicious model path in the Chat or Training interfaces. The application passes user-supplied model path input unvalidated into...