deepset-ai farm Model Package Loader language_model.py WordEmbedding_LM.load path traversal
Affects FARM at git commit 5919538f721c7974ea951b322d30a3c0e84a1bc2 (version string 0.8.1-snapshot) and the PyPI farm 0.8.0 wheel, when a model package uses WordEmbedding_LM. CVSS rates it 2.4 (low). No fix is listed.
Written by anthropic/claude-haiku-5.5 from the sources below. Verify against them.
- Published
- Last modified
- Where the record stands: published, rejected by its publisher, or superseded by another id.What do these mean? →
- Published
- Whether CISA lists it in its Known Exploited Vulnerabilities catalog: proof that attackers are using it.What do these mean? →
- Not listed in CISA KEV
- A 0-100 score for how much a CVE matters across the industry: exploitation, exposure, how widely the software is used, severity and whether a fix exists.What do these mean? →
- 12/100 · provisional
- How many real-world setups can hit the bug, from the vendor's advisory and CISA data. It does not know your setup.What do these mean? →
- Specific setups only
- Whether working attack code is public (Exploit-DB, Metasploit, Nuclei or GitHub). Public code makes attacks easier.What do these mean? →
- None on Exploit-DB, Metasploit or GitHub
Summary
The deepset-ai FARM framework (its Model Package Loader component) is affected by a path traversal flaw in how it handles the embeddings filename when loading word embedding models. A local attacker who can influence that input could read or access files outside the intended directory.
Read the original description
A vulnerability has been found in deepset-ai farm up to 0.8.0. Affected is the function WordEmbedding_LM.load of the file farm/modeling/language_model.py of the component Model Package Loader. The manipulation of the argument embeddings_filename leads to path traversal. An attack has to be approached locally. The exploit has been disclosed to the public and may be used. The vendor was contacted early about this disclosure but did not respond in any way. This vulnerability only affects products that are no longer supported by the maintainer.
What to do
Check whether your setup matches what the bug needs.
Affects deepset-ai FARM at git commit 5919538f721c7974ea951b322d30a3c0e84a1bc2 (version string 0.8.1-snapshot). The same vulnerable code is present in the PyPI farm wheel 0.8.0 and in the final public master revision. The model package must use the WordEmbedding_LM class and be loaded through LanguageModel.load().Workaround: The facts list no workaround. The flaw depends on the WordEmbedding_LM class being used in a loaded model package, so limiting which model packages are loaded is the only scoping the facts support.
Timeline
- CVE published
- First seen by cve.now
Affected versions
- deepset-ai farm
0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8.0
Evidence 5
- Advisory
- Advisory
- Advisory
- Advisory
- Exploit
Corrections
None so far.
Something wrong on this page? Report an error. Every fix is listed here.
Questions about CVE-2026-120335
What does this mean?
An attacker-supplied model package can make FARM read a file outside the package directory and convert its contents into model embedding weights. That can leak external data through the loaded model or poison the model.Source: github.com
Am I affected, and how do I check?
The advisory names FARM at git commit 5919538f721c7974ea951b322d30a3c0e84a1bc2 (version string 0.8.1-snapshot), and says the same vulnerable code is in the PyPI farm wheel 0.8.0 and the final public master revision. The flaw applies when a model package uses the WordEmbedding_LM class and is loaded through LanguageModel.load().Source: github.com
What should I do about it?
The facts list no fixed version, so no upgrade target can be named. Check for a later FARM release or advisory update before relying on any version.Source: github.com
Is it being exploited?
It is not in CISA's Known Exploited Vulnerabilities catalog, and CISA's SSVC assessment does not mark exploitation as active. That does not mean it has not been attacked: vendors and researchers can report attacks before CISA lists them. No public exploit code is tracked.Source: cve.now feed data
What if there's no patch?
The facts list no workaround. The flaw depends on the WordEmbedding_LM class being used in a loaded model package, so limiting which model packages are loaded is the only scoping the facts support.Source: github.com