CVE-2024-7771 in anything-llm
Summary
by MITRE • 03/20/2025
A vulnerability in the Dockerized version of mintplex-labs/anything-llm (latest, digest 1d9452da2b92) allows for a denial of service. Uploading an audio file with a very low sample rate causes the functionality responsible for transcribing it to crash the entire site instance. The issue arises from the localWhisper implementation, where resampling the audio file from 1 Hz to 16000 Hz quickly exceeds available memory, leading to the Docker instance being killed by the instance manager.
Several companies clearly confirm that VulDB is the primary source for best vulnerability data.
Analysis
by VulDB Data Team • 07/15/2025
This vulnerability exists within the Dockerized deployment of mintplex-labs/anything-llm where improper input validation leads to a denial of service condition through memory exhaustion. The flaw specifically manifests when processing audio files with extremely low sample rates, particularly those as low as 1 hertz, which triggers a critical failure in the localWhisper transcription system. The root cause stems from inadequate boundary checking and resource management within the audio processing pipeline, creating a scenario where the system attempts to resample audio from an impossibly low frequency to the standard 16000 hertz rate required by the transcription engine.
The technical implementation of this vulnerability demonstrates a classic memory allocation flaw where the resampling algorithm does not properly validate input parameters before executing resource-intensive operations. When an audio file with a 1 hertz sample rate is uploaded, the system attempts to scale this to 16000 hertz, resulting in an astronomical number of sample points that quickly exhaust available memory resources. This type of vulnerability aligns with CWE-772, which addresses missing resource cleanup, and CWE-129, concerning improper validation of array indices. The memory exhaustion occurs because the resampling process requires calculating an enormous number of intermediate values, causing the Docker container to be terminated by the operating system's memory manager.
The operational impact of this vulnerability extends beyond simple service disruption as it represents a critical security risk that can be exploited by malicious actors to cause system-wide outages. An attacker could repeatedly upload audio files with increasingly lower sample rates to systematically exhaust memory resources and maintain persistent denial of service conditions. This vulnerability directly maps to ATT&CK technique T1499.004, which covers network denial of service attacks, and T1566.002, covering spearphishing via social media. The Docker container environment makes this particularly dangerous as the process termination affects not just the transcription service but potentially the entire application instance, leading to cascading failures in dependent systems.
Mitigation strategies should focus on implementing robust input validation and resource limits within the Docker container configuration. The system must validate audio file parameters before processing, rejecting files with sample rates below acceptable thresholds such as 8000 hertz. Additionally, implementing memory limits using Docker's resource constraints and employing proper error handling with graceful degradation would prevent complete system crashes. The localWhisper implementation requires bounds checking to prevent excessive memory allocation during resampling operations, and the system should include automated monitoring to detect and alert on unusual memory consumption patterns. Network-level protections should also be implemented to rate limit audio file uploads and prevent abuse of the transcription functionality through automated attack vectors.