Attack ChainAIDerived
Hypothetical attack flow derived from CVE metadata – not a confirmed exploit chain
-
AccessGain local low-privilege access to Gradio host
-
DeliveryIdentify or craft two audio arrays with identical raw bytes but differing metadata
-
ExploitSubmit first audio to populate cache via save_audio_to_cache
-
ExecutionSubmit second audio triggering cache key collision
-
PersistRetrieve mismatched cached audio file
-
ImpactDisclose prior user's cached audio content
Vulnerability Assessment
| Weakness |
CWE
OWASP Top 10:2025
A04:2025 - Cryptographic Failures
ASVS 5.0
V11 - Cryptography Approved algorithm selection
|
| Exploitation Conditions | Exploitation requires local access to the system running Gradio (AV:L) with at least low-privilege credentials (PR:L per CVSS vector). High attack complexity (AC:H) means the attacker must precisely craft two audio arrays that share identical raw byte content while differing in at least one metadata attribute (sample rate, format, dtype, or shape) to produce a meaningful cache collision. Gradio's audio caching functionality must be active, which is the default behavior when the Audio component is used. Remote network exploitation is not possible. The limiting factors - mandatory local access, high complexity, and confidentiality-only impact limited to cached audio data - significantly constrain the realistic attacker population to co-located or shared-host scenarios involving mutually untrusted users. |
| Impact | Real-world risk is very low and the CVSS 4.0 score of 1.1 is consistent with the actual threat profile. The attack vector is strictly local (AV:L), requires low privileges (PR:L), and carries high attack complexity (AC:H), meaning the attacker must already have a foothold on the host and be able to craft precise audio inputs that produce a byte-level collision. The sole impact is limited confidentiality disclosure at the victim component level (VC:L) with no integrity or availability consequences. No EPSS data was provided in the source intelligence. The E:P exploit modifier confirms public proof-of-concept code exists, but no CISA KEV listing indicates no confirmed in-the-wild active exploitation. This vulnerability is a genuine finding with a narrow, well-defined blast radius - not a candidate for emergency prioritization unless Gradio instances are shared among mutually untrusted local users who handle sensitive audio data. |
| Exploit ScenarioDerived | A local user with low privileges submits two audio inputs to a Gradio application: one with dtype int16 at 8000 Hz and another with dtype uint16 at 8000 Hz, both containing identical raw byte sequences. Under the vulnerable hash scheme, both map to the same cache directory path, causing the second write to overwrite or collide with the first. A subsequent request for the first audio file retrieves the second user's cached audio instead, constituting information disclosure. A public proof-of-concept exists per the CVSS E:P modifier and the CVE description statement that 'the exploit has been released to the public.' |
Get an e-mail when this CVE is exploited, patched or re-scored. Free account.
Alert meAnalysisAIv1 – Jun 04, 2026
Cache key collision in Gradio 6.14.0's audio processing component allows a local low-privileged attacker to trigger information disclosure by exploiting incomplete hash inputs in the save_audio_to_cache function. Two audio arrays with identical raw bytes but differing metadata (sample rate, format, dtype, or shape) resolve to the same cache directory path, causing one cached audio file to overwrite or be served in place of another. Publicly available exploit code exists per the CVSS 4.0 E:P modifier and CVE description, though no active exploitation has been confirmed via CISA KEV. The CVSS 4.0 score of 1.1 accurately reflects the narrow real-world impact given the mandatory local access and high attack complexity.
RemediationAI
Apply the upstream fix from GitHub pull request #13394 (https://github.com/gradio-app/gradio/pull/13394), which replaces the incomplete hash with SHA-256 over a full metadata envelope. A specific patched release version was not confirmed in the provided intelligence beyond the changeset designation; monitor the official Gradio releases at https://github.com/gradio-app/gradio/ for the release incorporating this changeset. As a compensating control prior to patching, restrict local system access to the Gradio host so that only trusted users can interact with the application - this directly addresses the AV:L prerequisite. Disabling audio caching at the application level, if the feature is configurable, would also eliminate the collision surface entirely. Network-layer controls provide no mitigation benefit given the local-only attack vector.
Affected VersionsENISA
Affected Packages
Source: OSV (Open Source Vulnerabilities)
Affects your stack
Affects 1 package in pypi
Checking this CVE against your own stacks and servers is part of the Basic plan and up.
Technical ContextAI · show details
The vulnerability resides in gradio/processing_utils.py within the save_audio_to_cache function. Prior to fix PR #13394, cache directory paths were derived exclusively from hash_bytes(data.tobytes()) - hashing only the raw PCM byte content of the numpy audio array. Because the hash excluded metadata such as sample rate, format string, data type (e.g., int16 vs uint16), and array shape (mono vs stereo), two semantically distinct audio files with coincidentally identical raw bytes would hash to the same cache key. CWE-327 (Use of a Broken or Risky Cryptographic Algorithm) identifies the root cause class: the legacy hash function is cryptographically insufficient for its purpose as a unique cache discriminator. The fix in PR #13394 replaces this with SHA-256 over a JSON-serialized metadata envelope (cache_schema, dtype, format, sample_rate, shape) concatenated with the raw audio bytes, eliminating the collision vector entirely.
Know when Gradio gets its next vulnerability
- Watch products and get an email when something new is published
- Check this CVE against your own stack and servers
- Exploit and PoC references, threat intelligence from 22 sources
No credit card · 7-day full trial
Severity by source
CVSS:4.0/AV:L/AC:H/AT:N/PR:L/UI:N/VC:L/VI:N/VA:N/SC:N/SI:N/SA:N/E:P
Primary rating from Vendor (vuldb) · only source for this CVE.
CVSS VectorVendor: vuldb
DescriptionCVE.org
A security flaw has been discovered in gradio-app gradio 6.14.0. This affects the function save_audio_to_cache of the component Audio Cache Key Handler. Performing a manipulation results in use of weak hash. The attack must be initiated from a local position. The attack is considered to have high complexity. It is indicated that the exploitability is difficult. The exploit has been released to the public and may be used for attacks. The patch is named 13394. To fix this issue, it is recommended to deploy a patch.
Threat Intelligence Overview
How this decision was reached
The engine reads each signal below and applies a fixed list of rules from top to bottom. The first rule that matches sets the recommended action – like a statement, every line is shown.
What every action level means
SSVC AssessmentENISA
Priority Score
Attack Technique
Information disclosure occurs when an application unintentionally exposes sensitive data that aids attackers in reconnaissance or directly compromises security.
Gradio v4.36.1 was discovered to contain a code injection vulnerability via the component /gradio/component_meta.py. Rat
A local file include could be remotely triggered in Gradio due to a vulnerable user-supplied JSON value in an API reques
A command injection vulnerability exists in the gradio-app/gradio repository, specifically within the 'test-functional.y
A Server-Side Request Forgery (SSRF) vulnerability exists in the gradio-app/gradio version 4.21.0, specifically within t
A path traversal vulnerability exists in the Gradio Audio component of gradio-app/gradio, as of version git 98cbcae. Rat
A command injection vulnerability exists in the deploy+test-visual.yml workflow of the gradio-app/gradio repository, due
Command Injection in GitHub repository gradio-app/gradio prior to main. Rated high severity (CVSS 8.1), this vulnerabili
Gradio is an open source framework for building interactive machine learning models and demos. Rated high severity (CVSS
A Regular Expression Denial of Service (ReDoS) vulnerability exists in the gradio-app/gradio repository, affecting the g
A Denial of Service (DoS) vulnerability was discovered in the file upload feature of gradio-app/gradio version 0.39.1. R
A vulnerability in the file upload process of gradio-app/gradio version @gradio/video@0.10.2 allows for a Denial of Serv
A vulnerability in the dataframe component of gradio-app/gradio (version git 98cbcae) allows for a zip bomb attack. Rate
EUVD-2026-34187
GHSA-6655-8ph2-63j3
NVD
CVE.org
EPSS
Vulnrichment
VulDB
Exploit-DB