Code execution via unguarded pickle model loading
Published Dec 19, 2023 · Updated Aug 2, 2024
Unsafe deserialization in Hugging Face Transformers before 4.36.0 allows remote attackers to execute code through a malicious model. TransfoXLTokenizer and RagRetriever pass supplied vocabulary, passage, and index-metadata pickle files directly to pickle.load without a trust gate. Execution occurs when an application loads an attacker-controlled TransfoXL checkpoint or RAG index, and the payload runs with that application's privileges.
Summary
What happened
Unsafe deserialization in Hugging Face Transformers before 4.36.0 allows remote attackers to execute code through a malicious model. TransfoXLTokenizer and RagRetriever pass supplied vocabulary, passage, and index-metadata pickle files directly to pickle.load without a trust gate. Execution occurs when an application loads an attacker-controlled TransfoXL checkpoint or RAG index, and the payload runs with that application's privileges.
The record
- CVE
- CVE-2023-6730
- Published
- Dec 19, 2023
- Updated
- Aug 2, 2024
- Vendor
- Hugging Face
- Product
- Hugging Face Transformers
- Classifications
- CWE-502, T1204.002
- Attack vector
- network
- Privileges
- authenticated
Timeline
How it unfolded
- Dec 19, 2023CVE publishedPublication date reported by the CVE source.
- Aug 2, 2024Record updatedLatest update available in the CVE record.
Exploitability
Present is not the same as exploitable
Compare your product and version with the public record. A matching version still requires validation against your environment.
Is a vulnerable build present?
Compare these published version ranges with your installed build and any vendor patches.
- Affected versionversion=unspecified <4.36
What conditions does exploitation require?
What is affected?
Attacks
What attackers are doing with it
Daily unique IPs observed by Shadowserver honeypots for known exploited vulnerabilities (KEVs). Missing observations do not establish an absence of attacks.
Weakness, pattern, technique
Public exploit references
- Huntr malicious RAG model exploitproof of concept · unverified
Labels summarize the accepted research assessment. They do not indicate a test against your environment.
Technologies
Your stack
See the directory against your own environment.
Your stack
Check the software in your environment
Book a demo to see how Hinoki identifies affected software and validates exploitability in your environment.
Book a demo