IBM watsonx.data EUVD-2025-209603

Requires a deployment of IBM watsonx.data 2.2.0 through 2.3.0 (IBM Lakehouse) running on its containerized/Kubernetes-OpenShift pod architecture, where network policies between pods are not enforced. Exploitation realistically requires the attacker to already have a position on the cluster/pod network (e.g., a compromised or attacker-controlled pod) to abuse the unrestricted pod-to-pod communication channel (CWE-923). Limiting factors: the flaw is a segmentation gap rather than a direct code-execution or data-read primitive; impact is confined to unauthorized data transfer between pods (I:H, with C:N/A:N); no user interaction is needed, but the practical need for in-cluster presence meaningfully narrows the exposure versus the AV:N rating.

This is a high-CVSS (7.5) but low-real-world-urgency issue when all signals are weighed together. The vendor CVSS vector (AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:H/A:N) implies a trivially exploitable network flaw, but the EPSS score is only 0.01% (3rd percentile) and the SSVC decision framework rates Exploitation as 'none', Automatable as 'no', and Technical Impact as 'partial' - all pointing to a low-priority, non-weaponized weakness. There is no CISA KEV listing and no public exploit identified at time of analysis. The CVSS vector itself is internally notable: it scores Integrity-High (I:H) with Confidentiality and Availability at None, yet the product Tags label this 'Information Disclosure' - these conflict, and the vector (data can be transferred/moved between pods, an integrity/lateral-movement concern) is better aligned to integrity than to disclosure. Also worth verifying: the AV:N/PR:N rating treats this as fully remote-unauthenticated, but the described weakness (CWE-923, improper restriction of communication between Kubernetes/OpenShift pods) realistically presupposes an attacker already positioned inside the cluster or on an adjacent pod network - verify the intended attack position with IBM. Net: a legitimate hardening/segmentation defect to patch on a normal cycle, not an emergency.

An attacker who has already gained a foothold in one workload pod within the watsonx.data / IBM Lakehouse cluster leverages the missing inter-pod communication restrictions to reach and exchange data with other pods that should have been network-isolated. Because segmentation is not enforced, the attacker moves or exfiltrates data laterally between services without authorization. No public proof-of-concept exists, and the realistic prerequisite is existing access to the cluster network rather than pure internet-facing reach.