Serialization Injection Vulnerability in LangChain Framework
CVE-2025-68664

9.3CRITICAL

Key Information:

Vendor: Langchain-ai
Status: Langchain
Vendor: CVE Published:; 23 December 2025

🔔 Create Langchain-ai Vulnerability Alert

Badges

📈 Trended📈 Score: 1,470👾 Exploit Exists📰 News Worthy

What is CVE-2025-68664?

CVE-2025-68664 is a serialization injection vulnerability found in the LangChain framework, which is designed for creating agents and applications powered by large language models (LLMs). This vulnerability arises from the improper handling of user data within the dumps() and dumpd() functions, where dictionaries containing the 'lc' key are not escaped properly during serialization. The 'lc' key serves an internal purpose within LangChain to identify serialized objects, but when user-controlled data incorporates this structure, it could be misinterpreted during deserialization as a legitimate object rather than benign user data. Such a flaw could allow malicious actors to manipulate or inject harmful data into applications built on the LangChain framework, compromising the integrity and security of the entire platform.

Potential impact of CVE-2025-68664

Data Manipulation: Attackers could exploit this vulnerability to inject malicious data that the LangChain applications might process as valid, leading to unintended outcomes and potential application-level compromises.
Security Breach Risk: Given the nature of the vulnerability, an attacker could potentially gain unauthorized access or control over applications using the affected functions, heightening the risk of data breaches and unauthorized operations.
Reputational Damage: Organizations relying on the LangChain framework could suffer significant reputational harm if the vulnerability is exploited, particularly if sensitive user data is compromised or if application functionality is disrupted due to malicious activity.

Human OS v1.0:
Ageing Is an Unpatched Zero-Day Vulnerability.

Remediate biological technical debt. Prime Ageing uses 95% high-purity SIRT6 activation to maintain genomic integrity and bolster systemic resilience.

Deploy the Patch

Affected Version(s)

langchain >= 1.0.0, < 1.2.5 < 1.0.0, 1.2.5

langchain < 0.3.81 < 0.3.81

News Articles

prismedia.ai

CVE-2025-68664

Critical LangChain "LangGrinch" Flaw Exposes Secrets - Patch Now

A critical serialization vulnerability in langchain-core, tracked as CVE-2025-68664 and nicknamed LangGrinch, can allow attackers to exfiltrate API keys, manipulate LLM outputs, and in some cases trigger remote code execution. Security teams and developers should update affected packages immediately...