high CVSS 7.5

CVE-2026-41523·Python vulnerability

vLLM is an inference and serving engine for large language models (LLMs). Prior to 0.22.0, an assert-based security check in vLLM's activation function loading allows any unauthenticated attacker to achieve arbitrary code execution on the server by publishing a malicious HuggingFace model, when vLLM runs in Python optimized mode (python -O or PYTHONOPTIMIZE=1). This vulnerability is fixed in 0.22.0.

Severity
high
Software
Python
Fixed in
0.22.0
Published
2026-06-22

Affected versions

Until: 0.22.0

Fixed in: 0.22.0

How to fix this CVE

Upgrade vLLM to version 0.22.0 or later to eliminate the assertion-based security check bypass that permits arbitrary code execution when running Python in optimized mode. Organizations running vLLM with PYTHONOPTIMIZE=1 or python -O flags should prioritize this update immediately, as the vulnerability allows remote attackers to execute malicious code by distributing compromised HuggingFace models. Disable Python optimization flags in production environments until the patch is deployed.

sudo dnf update python3 python3-vllm

Defensia detects this vulnerability

How to check if you are affected

  1. Step 1: Check vLLM version with: python3 -c "import vllm; print(vllm.__version__)"
  2. Step 2: Verify Python optimization is disabled by running: python3 -c "import sys; print('Optimization enabled' if sys.flags.optimize > 0 else 'Optimization disabled')"
  3. Step 3: Review process startup arguments with: ps aux | grep -i vllm to confirm no -O flag or PYTHONOPTIMIZE environment variable is set
  4. Step 4: Confirm patch application by verifying vLLM version is 0.22.0 or later and re-run activation function loading without assertion bypass

FAQ

What is CVE-2026-41523?

CVE-2026-41523 is a critical code execution vulnerability in vLLM where disabled assertion checks in Python optimized mode allow unauthenticated attackers to execute arbitrary code by hosting malicious HuggingFace models that vLLM loads during inference.

Is CVE-2026-41523 being actively exploited?

No, this vulnerability is not currently listed in the CISA Known Exploited Vulnerabilities (KEV) database and no public exploits have been released, though the attack surface is significant for organizations running vLLM with Python optimization enabled.

What versions of vLLM and Python are affected by CVE-2026-41523?

All vLLM versions prior to 0.22.0 are vulnerable when executed with Python's optimization flags (-O or PYTHONOPTIMIZE=1 environment variable), regardless of Python 3.x version.

How do I check if my server is vulnerable to CVE-2026-41523?

Run: python3 -c "import vllm; print(vllm.__version__)" and check if the version is below 0.22.0, then verify Python optimization with: python3 -c "import sys; print(sys.flags.optimize)". A version below 0.22.0 AND optimization flag > 0 indicates vulnerability.

Does Defensia detect CVE-2026-41523?

Yes — Defensia's CVE advisory scanner compares installed package versions against the NVD database. If vLLM is installed on a monitored server, CVE-2026-41523 will appear in your dashboard with remediation steps and configuration risk assessment.

Related Python CVEs

CVE-2026-25632CVSS 10EPyT-Flow is a Python package designed for the easy generation of hydraulic and water quality scenario data of water distribution networks. Prior to 0.16.1, EPyT-Flow’s REST API parses attacker-controlled JSON request bodies using a custom deserializer (my_load_from_json) that supports a type field. When type is present, the deserializer dynamically imports an attacker-specified module/class and instantiates it with attacker-supplied arguments. This allows invoking dangerous classes such as subprocess.Popen, which can lead to OS command execution during JSON parsing. This also affects the loading of JSON files. This vulnerability is fixed in 0.16.1.
CVE-2026-33054CVSS 10Mesop is a Python-based UI framework that allows users to build web applications. Versions 1.2.2 and below contain a Path Traversal vulnerability that allows any user supplying an untrusted state_token through the UI stream payload to arbitrarily target files on the disk under the standard file-based runtime backend. This can result in application denial of service (via crash loops when reading non-msgpack target files as configurations), or arbitrary file manipulation. This vulnerability heavily exposes systems hosted utilizing FileStateSessionBackend. Unauthorized malicious actors could interact with arbitrary payloads overwriting or explicitly removing underlying service resources natively outside the application bounds. This issue has been fixed in version 1.2.3.
CVE-2026-34938CVSS 10PraisonAI is a multi-agent teams system. Prior to version 1.5.90, execute_code() in praisonai-agents runs attacker-controlled Python inside a three-layer sandbox that can be fully bypassed by passing a str subclass with an overridden startswith() method to the _safe_getattr wrapper, achieving arbitrary OS command execution on the host. This issue has been patched in version 1.5.90.
CVE-2026-28505CVSS 10Tautulli is a Python based monitoring and tracking tool for Plex Media Server. Prior to version 2.17.0, the str_eval() function in notification_handler.py implements a sandboxed eval() for notification text templates. The sandbox attempts to restrict callable names by inspecting code.co_names of the compiled code object. However, co_names only contains names from the outer code object. When a lambda expression is used, it creates a nested code object whose attribute accesses are stored in code.co_consts, NOT in code.co_names. The sandbox never inspects nested code objects. This issue has been patched in version 2.17.0.
CVE-2026-25115CVSS 9.9n8n is an open source workflow automation platform. Prior to version 2.4.8, a vulnerability in the Python Code node allows authenticated users to break out of the Python sandbox environment and execute code outside the intended security boundary. This issue has been patched in version 2.4.8.

References

Track CVEs across your fleet automatically

Defensia scans your Linux servers and tells you exactly which ones are running vulnerable versions — including CVE-2026-41523. Free for 1 server.

Get started free