AI Sec Watch

LangChain Experimental versions 0.1.17 through 0.3.0 contain a vulnerability that allows attackers to execute arbitrary code (run malicious commands on a system) through a component called LLMSymbolicMathChain, which uses sympy.sympify (a function that evaluates mathematical expressions in an unsafe way). The root cause is improper input validation (failing to check that user input is safe before processing it).

A vulnerability in the ilab model serve component allows attackers to cause a Denial of Service (DoS, where a service becomes unavailable to legitimate users) by sending a large value for the best_of parameter to the vllm JSON web API (a web interface for accessing an LLM). The API doesn't properly manage timeouts or resource limits, so an attacker can exhaust system resources and crash the service.

CVE-2024-8768 is a bug in vLLM (a library for running large language models) where sending an API request with an empty prompt crashes the server, causing a denial of service (making the service unavailable to users). The flaw is classified as a reachable assertion vulnerability, meaning the code hits an unexpected condition it wasn't designed to handle.

A vulnerability in langchain's FAISS.deserialize_from_bytes function allows deserialization of untrusted data using pickle (a Python library that converts data into a format that can be stored or transmitted), which can lead to arbitrary command execution through the os.system function. This affects the latest version of the product and is classified as CWE-502 (deserialization of untrusted data).

CVE-2024-6587 is a server-side request forgery vulnerability (SSRF, a flaw that tricks a server into making requests to unintended locations) in litellm version 1.38.10 that lets users control where the application sends requests by setting the `api_base` parameter, potentially allowing attackers to intercept sensitive OpenAI API keys. A malicious user could redirect requests to their own domain and steal the API key, gaining unauthorized access to the OpenAI service.

MindsDB versions 23.12.4.0 through 24.7.4.1 contain an arbitrary code execution vulnerability (the ability to run unwanted commands on a server) when the ChromaDB integration is installed. An attacker can craft a malicious 'INSERT' query containing Python code that gets executed on the server because the code is passed to an eval function (a function that runs text as if it were code).

MindsDB versions 23.10.3.0 through 24.7.4.1 have a vulnerability that allows arbitrary code execution (running unauthorized commands on a server) when the Weaviate integration is installed. An attacker can exploit this by crafting a malicious SQL SELECT WHERE clause containing Python code, which gets executed through an eval function (a function that interprets and runs code as if it were written in the program).

CVE-2024-45855 is a vulnerability in MindsDB (a platform for building AI applications) versions 23.10.2.0 and newer where deserialization of untrusted data (converting data from an external format into code without checking if it's safe) can occur. An attacker can upload a malicious 'inhouse' model and use the 'finetune' feature to run arbitrary code (any commands they want) on the server.

CVE-2024-45854 is a vulnerability in MindsDB (a platform for building AI applications) versions 23.10.3.0 and newer where deserialization of untrusted data (converting data from an external format back into executable code without checking if it's safe) allows an attacker to upload a malicious model that runs arbitrary code (any commands the attacker wants) on the server when a describe query is executed on it.