CVE-2026-7669
Official description Straight from the sourceThe vendor's or NVD's own wording, published unedited. Authoritative, but often terse — it says what broke, rarely what to do.
NVD · uneditedA vulnerability was detected in sgl-project SGLang up to 0.5.9. Impacted is the function get_tokenizer of the file python/sglang/srt/utils/hf_transformers_utils.py of the component HuggingFace Transformer Handler. The manipulation of the argument trust_remote_code with the input False as part of Boolean results in code injection. The attack can be executed remotely. A high complexity level is associated with this attack. The exploitability is considered difficult. In get_tokenizer(), when the caller passes trust_remote_code=False and HuggingFace transformers v5 returns a TokenizersBackend instance (the generic fallback for tokenizer classes not in the registry), SGLang silently re-invokes AutoTokenizer.from_pretrained with trust_remote_code=True, overriding the caller's explicit security setting. A model repository containing a malicious tokenizer.py referenced via auto_map in tokenizer_config.json will execute arbitrary Python in the SGLang process during this second call. No log line or warning is emitted. The override affects all current SGLang versions because transformers==5.3.0 is pinned in pyproject.toml. Both tokenizer_mode="auto" and tokenizer_mode="slow" are affected. The exploit is now public and may be used. The vendor was contacted early about this disclosure but did not respond in any way.
Technical summary Written by usOur analysis, written from the advisory, the CVSS vector and the affected-version data. It adds context the advisory leaves out, and never invents facts that are not in the source.
dbcve analysis · high confidenceIn SGLang's get_tokenizer() function, when HuggingFace transformers v5 returns a TokenizersBackend fallback instance (for unregistered tokenizer classes), the code silently re-invokes AutoTokenizer.from_pretrained with trust_remote_code=True, overriding the caller's explicit False setting. This allows a malicious model repository with a crafted tokenizer.py in tokenizer_config.json to achieve arbitrary code execution in the SGLang process.
Verify against the referenced sources before acting — the references below are authoritative for this CVE, this summary is not.
CVSS breakdown How the score is builtThe industry scoring standard. It rates how the flaw is reached, what it takes to exploit, and what an attacker gains — the score is derived from those, not the other way round.
From the vector- Attack vector
- Network
- Complexity
- High
- Privileges
- None
- User interaction
- None
- Scope
- Unchanged
- Confidentiality
- Low
- Integrity
- Low
- Availability
- Low
CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:L/I:L/A:L
Am I affected? How to checkSteps we derive from the advisory and the affected-version data, so you can decide whether this CVE reaches your setup. They are a guide, not a scan — your own configuration is the authority.
dbcve checksWork through these to decide whether this CVE applies to you.
-
Identify SGLang installationRun 'pip show sglang' or 'python -c "import sglang; print(sglang.__file__)"' to locate the SGLang package installationAffected if SGLang is installed and the vulnerable get_tokenizer() function exists in the codebase
-
Locate the get_tokenizer function sourceFind the file containing get_tokenizer() in the SGLang package, typically in the sglang python package directoryAffected if The function is found and can be inspected for the vulnerable code pattern
-
Inspect get_tokenizer for trust_remote_code overrideOpen the get_tokenizer source file and search for code that re-invokes AutoTokenizer.from_pretrained with trust_remote_code=True after receiving a TokenizersBackend fallbackAffected if The code explicitly sets trust_remote_code=True when loading a fallback tokenizer, overriding the caller's False setting
-
Check if tokenizer loading uses HuggingFace modelsReview application code or configuration that calls get_tokenizer() to see if it loads tokenizers from HuggingFace Hub or other remote repositoriesAffected if The application loads tokenizers from remote model repositories using this function
-
Verify transformers versionRun 'pip show transformers' to check if HuggingFace transformers v5 is installed, which triggers the TokenizersBackend fallback behaviorAffected if Transformers version 5.x is installed and the fallback code path is reachable
A user is affected if SGLang is installed with the vulnerable get_tokenizer() function that overrides trust_remote_code=False to True, and the system loads tokenizers from remote HuggingFace model repositories.
Generated from the published advisory. Verify against your own configuration.
Remediation Closing itWhat it takes to close this. Where a vendor fix exists we point at it; where none exists we say so plainly, and can build one. Effort estimates are scoped from the advisory, not from your codebase.
From vendor dataModify get_tokenizer() to respect the caller's trust_remote_code=False setting instead of overriding it; do not silently re-invoke with trust_remote_code=True. Alternatively, log a warning and refuse to load the tokenizer rather than bypass the security setting.
- Consultation2.0 h
- Implementation3.0 h
- Testing3.0 h
- Review / QA2.0 h
An estimate, not a bill — we confirm scope with you before any work starts. Need it this week? Rush from $2,800.
Scan for this in your stack
Free · runs locallyCheck whether your project pulls in CVE-2026-7669 — or any other known-vulnerable package — straight from your lock files. Free and open source; it runs locally and uploads nothing.
References Go to the primary sourcePrimary sources — vendor advisories, patches and trackers. Where our summary and a reference disagree, the reference wins.
Primary sourcesPractitioner notes
ContributedPeer-ranked notes from engineers who’ve handled CVE-2026-7669 in production — separate from our analysis above.
The advisory tells you what broke. It rarely tells you what actually worked. If you’ve dealt with this one, that detail is what the next engineer is searching for.
- The version that genuinely resolved it — not the one the vendor claimed
- A config change or rule that shut the vector down
- A gotcha in the upgrade path that cost you an afternoon
No notes yet
Be the first to add a field note for this CVE — a mitigation you’ve verified, a version caveat, or a link to a working fix. Sign in above to contribute.
A place for practitioners to share what actually worked: a mitigation you’ve tested, a configuration change, a version- or environment-specific caveat, or a link to a verified patch. The most useful notes rise to the top as peers upvote them, so the signal stays high.
- Verified mitigations, workarounds, and config changes
- Version or environment caveats, and links to real fixes
- No weaponised exploit code, or anything meant to cause harm
- No spam, self-promotion, credentials, or personal data