vantage6
CVE-2026-54533
MEDIUM
Severity by source
CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:N/VI:L/VA:N/SC:N/SI:N/SA:N/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X
Network-reachable via algorithm submission with no authentication (per PR:N tag and 'Authentication Bypass'); high confidentiality impact because private analysis data of co-resident algorithms is exposed.
Primary rating from Vendor (GitHub_M).
CVSS VectorVendor: GitHub_M
CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:N/VI:L/VA:N/SC:N/SI:N/SA:N/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X
Lifecycle Timeline
2Blast Radius
ecosystem impact- 3 pypi packages depend on vantage6 (3 direct, 0 indirect)
Ecosystem-wide dependent count for version 5.0.0.
DescriptionCVE.org
vantage6 is an open-source infrastructure for privacy preserving analysis. Prior to version 5.0.0, malicious algorithms can potentially access other algorithms input and output files. Version 5.0.0 fixes the issue. As a workaround, verify and restrict the algorithm containers that are allowed to run on the node.
AnalysisAI
Improper access control in vantage6 nodes prior to version 5.0.0 allows malicious algorithm containers to read input and output files belonging to other algorithms running on the same node. This directly undermines the core privacy guarantee of the platform - a federated learning infrastructure explicitly designed for privacy-preserving analysis - by exposing sensitive intermediate data to adversarially crafted algorithms. No public exploit has been identified at time of analysis, and a patch is available in version 5.0.0.
Technical ContextAI
vantage6 (cpe:2.3:a:vantage6:vantage6:*:*:*:*:*:*:*:*) is an open-source Python-based federated learning framework that orchestrates algorithm execution across distributed nodes, each of which runs algorithm containers (Docker-based) on behalf of researchers. The vulnerability is classified as CWE-284 (Improper Access Control): the node runtime fails to enforce filesystem isolation between algorithm containers, allowing one container to traverse or read the working directories of co-resident algorithm containers. In federated learning, node input files typically contain partitioned sensitive datasets (e.g., clinical records, financial transactions) and output files contain intermediate model parameters or query results - all of which are meant to remain opaque to other algorithms. The failure to sandbox container file access at the node level is the root cause.
RemediationAI
Vendor-released patch: vantage6 5.0.0. Upgrade node deployments to version 5.0.0 or later via pip (pip install --upgrade vantage6) and restart the node service. As a compensating control for nodes that cannot be immediately upgraded, restrict which algorithm images are permitted to execute on the node using the allowlist mechanism documented at https://docs.vantage6.ai/usage/running-the-node/security - this reduces the attack surface by ensuring only trusted, vetted algorithm containers are scheduled. Note that this workaround requires ongoing operational diligence: any gap in the allowlist policy reintroduces risk. For additional defense-in-depth, network-segment node hosts so that only authorized orchestrators can submit tasks, reducing the pool of parties able to submit malicious algorithms. Reference the full advisory at https://github.com/vantage6/vantage6/security/advisories/GHSA-x9f6-9rvm-mmrg.
vantage6 is privacy preserving federated learning infrastructure. Rated high severity (CVSS 7.2), this vulnerability is
Critical authentication bypass vulnerability in vantage6 (an open-source federated learning and privacy-enhancing techno
The vantage6 technology enables to manage and deploy privacy enhancing technologies like Federated Learning (FL) and Mul
The vantage6 technology enables to manage and deploy privacy enhancing technologies like Federated Learning (FL) and Mul
vantage6 is a framework to manage and deploy privacy enhancing technologies like Federated Learning (FL) and Multi-Party
vantage6 is a privacy preserving federated learning infrastructure for secure insight exchange. Rated high severity (CVS
vantage6 servers auto-generate JWT secret keys using UUID1, a predictable algorithm that lacks cryptographic strength, a
Broken authorization in vantage6's algorithm store allows any authenticated algorithm developer to overwrite a peer's pe
Default hardcoded admin credentials in vantage6 expose servers running versions prior to 5.0.0 to unauthorized administr
vantage6 is an open source framework built to enable, manage and deploy privacy enhancing technologies like Federated Le
vantage6 is a privacy preserving federated learning infrastructure for secure insight exchange. Rated medium severity (C
vantage6 is privacy preserving federated learning infrastructure. Rated medium severity (CVSS 5.4), this vulnerability i
Same weakness CWE-284 – Improper Access Control
View allSame technique Authentication Bypass
View allShare
External POC / Exploit Code
Leaving vuln.today