GPUStack: Missing authorization on several cluster and resource endpoints allows cross-user access and cluster takeover
Affects GPUStack 2.2.0 before 2.2.2 when a Kubernetes-provider cluster has a reachable worker. Fixed in 2.2.2.
Written by anthropic/claude-haiku-5.5 from the sources below. Verify against them.
- Exploited
- No known exploitation
- Fix
- Fixed in 2.2.2
- Who is exposed
- Specific setups only
Do first: Check whether your setup matches what the bug needs. Kubernetes-provider cluster with a reachable worker
- Exploitation?/40Exploitation: unknown (data gap). No CISA SSVC exploitation rating yet
- Exposure8/25Specific setups only, reachable over the network
- Reach?/20Reach: unknown (data gap). No tracked technology or package matched
- Severity8.8/10CVSS 8.8
- No patch0/5No signal
What would change the score
- Would score about 67 if added to CISA KEV
- Would score about 62 if CISA's SSVC rates exploitation active
- Would score about 24 if CISA's SSVC rates it automatable
CVE-2026-103002 is not in CISA's Known Exploited Vulnerabilities (KEV) catalog. No public exploit code is tracked. It is rated CVSS 8.8 (high) and affects gpustack; fixed in 2.2.2.
- Last modified
- Where the record stands: published, rejected by its publisher, or superseded by another id.What do these mean? →
- Published
- Whether CISA lists it in its Known Exploited Vulnerabilities catalog: proof that attackers are using it.What do these mean? →
- Not listed in CISA KEV
- A 0-100 score for how much a CVE matters across the industry: exploitation, exposure, how widely the software is used, severity and whether a fix exists.What do these mean? →
- 17/100 · provisional
- How many real-world setups can hit the bug, from the vendor's advisory and CISA data. It does not know your setup.What do these mean? →
- Specific setups only
- Fix
- Fixed in 2.2.2
- Whether working attack code is public (Exploit-DB, Metasploit, Nuclei or GitHub). Public code makes attacks easier.What do these mean? →
- None on Exploit-DB, Metasploit or GitHub
Summary
GPUStack, an open-source GPU cluster manager for AI model serving, has missing authorization checks on its cluster API proxy that affect versions 2.2.0 through 2.2.2. An authenticated low-privilege user can remotely reach Kubernetes clusters they should not access and read, change, or delete resources using a worker's ServiceAccount permissions, and this can lead to code execution in the target cluster when the cluster uses a Kubernetes provider with a reachable worker.
Read the original description
GPUStack is an open-source GPU cluster manager for AI model serving and GPU instance provisioning. From 2.2.0 until 2.2.2, cluster_apiserver_proxy forwarded an authenticated non-administrator's requests to a selected Kubernetes cluster without a resource-level authorization check. The proxy used a reachable worker's Kubernetes ServiceAccount credentials, so the caller could read sensitive resources, create or modify workloads, or delete resources according to that ServiceAccount's permissions. Exploitation required a Kubernetes-provider cluster with a reachable worker, and attacker-controlled workloads could result in code execution in the target cluster. This issue is fixed in version 2.2.2.
What to do
Check whether your setup matches what the bug needs.
Kubernetes-provider cluster with a reachable workerUpgrade to a fixed version: 2.2.2.
Workaround: Restrict access to the GPUStack instance to trusted administrator accounts until upgrading.
Timeline
- CVE published
- First seen by cve.now
Affected versions
- gpustack gpustack
From 2.2.0 · fixed in 2.2.2
Evidence 6
- Advisory
- Repository
- Repository
- Repository
- Repository
- Repository
Corrections
None so far.
Something wrong on this page? Report an error. Every fix is listed here.
Questions about CVE-2026-103002
Is CVE-2026-103002 being exploited?
It is not in CISA's Known Exploited Vulnerabilities catalog, and CISA's SSVC assessment does not mark exploitation as active. That does not mean it has not been attacked: vendors and researchers can report attacks before CISA lists them. No public exploit code is tracked.Source: cve.now feed data
What does this mean?
Several cluster and resource endpoints in GPUStack do not check authorization, so one user can reach resources that belong to others and take over a cluster. Depending on the permissions of the worker's ServiceAccount, the impact can include reading sensitive Kubernetes resources, creating or changing workloads, deleting resources, and running code in the target cluster.Source: github.com
Which versions are affected by CVE-2026-103002?
gpustack: from 2.2.0 before 2.2.2.Source: cve.now feed data
Am I affected, and how do I check?
GPUStack versions from 2.2.0 up to but not including 2.2.2 are in the affected range. The affected setup is a Kubernetes-provider cluster with a reachable worker, so check the installed GPUStack version and whether any cluster uses the Kubernetes provider with a reachable worker.Source: github.com
How do I fix CVE-2026-103002?
Upgrade GPUStack to 2.2.2, the fixed version. Until the upgrade is done, restrict access to the GPUStack instance to trusted administrator accounts.Source: github.com