Lighthouse AdvisorySLED AI Adoption Intelligence
← Back to results

From the Red Hat edition of November 9, 2026

Vendor claimCautionaryRecent

CVE-2026-83495: Low-impact Trainer v2 TrainJob secretKeyRef injection in OpenShift AI (Technology Preview) enables same-namespace secret exfiltration where RBAC is mismatched

Red Hat Product Security · Cross-sector technology vendor (OpenShift AI) · Global product

Red Hat
Publisher
Red Hat Customer Portal
Original publication
Red Hat public date 2026-08-31; CVE page last changed 2026-11-07 (UTC)
Source retrieved
2026-11-10
Event date
2026-11-07
Read original source

What happened

Red Hat rates CVE-2026-83495 Low because Trainer v2 is a Technology Preview feature of OpenShift AI. A user who can create TrainJobs can inject valueFrom.secretKeyRef into environment variables to read and exfiltrate namespace secrets, a privilege escalation when TrainJob RBAC is broader than Secret read RBAC. Mitigation: align TrainJob creation RBAC with Secret read permissions, or restrict secretKeyRef to an allowlist in the validating webhook. Package rhoai/odh-trainer-rhel9 is Under investigation with no advisory.

Why it matters

University labs and agency data-science teams often grant broad job-creation rights while restricting secret access; this flaw shows such a mismatch can expose credentials such as data-source keys within a project namespace.

Evidence and measured results

Opened Red Hat CVE page: description, Low statement citing Technology Preview, mitigation text, Bugzilla 2526350, CWE-269; record public date 2026-08-31, last changed 2026-11-07T06:02Z, package Under investigation.

Limitations and uncertainty

Low impact per vendor; affects a Technology Preview component; Under investigation; no CVSS score shown.

Cross-source patterns this evidence supports

Put this evidence to work

Lighthouse Advisory interpretation, grounded in this source. Enriched 2026-11-09; this does not change the original publication date. Each role follows the same checklist so takeaways compare across sources.

Sales

Role takeaway
Customer problem
Research and analytics namespaces may grant job-creation rights broader than secret access, enabling credential exposure.
Stakeholders
  • Research computing security lead
  • Data science platform owner
Discovery questions
  • Is Trainer v2 enabled anywhere?
  • Do job creators also hold secret-read rights?
Value hypothesis
Bounded: RBAC alignment closes the documented exposure path.
Potential engagement
An RBAC alignment review for OpenShift AI namespaces.
Unsupported claims
  • High severity
  • Cross-namespace impact
  • A fix timeline

Pre-sales engineering

Role takeaway
Fit
Relevant only where Trainer v2 Technology Preview is enabled.
Architecture and integration
Trainer v2 TrainJob API environment passthrough with secretKeyRef.
Prerequisites
  • RBAC inventory
  • Record of preview-feature enablement
Constraints
  • Technology Preview component
  • Package Under investigation
Security
Align RBAC or allowlist secretKeyRef in the webhook; use an external secret store.
Proof of value
  • Attempt secretKeyRef injection in staging after mitigation and confirm denial
  • Run an RBAC mismatch report

Delivery

Role takeaway
Work
Remediate RBAC mismatches for TrainJob creators.
Dependencies
  • RBAC tooling
  • Webhook configuration
Ownership
OpenShift AI platform team.
Skills and adoption
Platform admins need an RBAC review procedure.
Governance checkpoints
  • Decision record on Technology Preview enablement
  • RBAC change approval
Acceptance criteria
  • Proposed: zero namespaces where TrainJob create exceeds Secret read
  • Proposed: staging injection test denied
Risks
  • Breaking legitimate training jobs
  • Overlooked namespaces

Implementation considerations

Lighthouse Advisory interpretation across the operating dimensions a public-sector buyer must settle before this evidence becomes a design. Each note answers the question under its heading for this specific source.

Architecture and integration

What must connect, and where does the AI sit in the workflow?

Keep data-source secrets in separate namespaces or external secret stores, and avoid job-creation rights broader than secret-read rights.

Governance

Who approves, reviews and stays accountable for outcomes?

Technology Preview features such as Trainer v2 should stay out of production research platforms unless formally risk-accepted.

Security and privacy

What data, permissions and controls need testing?

Confidentiality risk for namespace secrets, limited to the same namespace.

Accessibility and workforce

Who is affected, and what skills or accommodations follow?

Not addressed in the source.

Procurement

What should contracts, pricing and exit terms secure?

Not material beyond standard vulnerability-response SLAs.

Operating model

Which teams own the service once it runs?

RBAC reviews should compare job-creation and secret-read permissions per namespace.

Publication history

  1. 2026-11-09Red Hat · Issue 107 resources
Read preserved resource versions (JSON)

Stable resource ID: rhoai-cve-2026-83495-trainer-v2-secretkeyref-exfiltration