EUVD-2026-42382
GHSA-jm96-7wvj-j3r9
Severity by source
AV:A/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H
Description references a 'public detection API' implying network reachability (AV:N over AV:A); no auth or complexity required; impact is availability-only.
Primary rating from Vendor (redhat).
CVSS VectorVendor: redhat
Lifecycle Timeline
1DescriptionCVE.org
A flaw was found in guardrails-detectors, a component of Red Hat OpenShift AI. This vulnerability, known as Regular Expression Denial of Service (ReDoS), allows a remote attacker to provide specially crafted regular expressions to the public detection API. This can cause catastrophic backtracking, leading to a worker process consuming 100% CPU indefinitely and resulting in a denial of service for the entire guardrails-mediated LLM pipeline.
AnalysisAI
ReDoS vulnerability in the guardrails-detectors component of Red Hat OpenShift AI enables adjacent-network attackers to submit specially crafted regular expressions to the public detection API, triggering catastrophic backtracking that pins a worker process at 100% CPU indefinitely. The availability impact extends beyond the component itself - the entire guardrails-mediated LLM pipeline is rendered non-functional. No public exploit identified at time of analysis, but exploitation requires no authentication and trivial effort from an adjacent network position.
Technical ContextAI
CWE-1333 (Inefficient Regular Expression Complexity) describes vulnerabilities where a regex engine exhibits exponential or polynomial time complexity against adversarially constructed patterns - the classic ReDoS class. The guardrails-detectors component of Red Hat OpenShift AI exposes a detection API that processes user-supplied or attacker-controlled regular expressions without sufficient complexity limits, timeouts, or input sanitization. When the underlying regex engine evaluates a malicious pattern containing deeply nested alternations or quantifiers, it exhausts all possible backtracking paths before determining non-match, causing indefinite CPU saturation. The affected component acts as a safety guardrail intermediary for LLM inference pipelines, meaning its unavailability cascades to all downstream AI workloads it mediates.
RemediationAI
Apply the vendor-released patch for the guardrails-detectors component of Red Hat OpenShift AI; consult https://access.redhat.com/security/cve/CVE-2026-15154 for the specific patched version and errata - exact fix version numbers are not independently confirmed from available data. As an immediate compensating control, restrict access to the detection API endpoint to only trusted, authenticated internal callers, reducing exposure from the adjacent network vector (trade-off: may require application changes or API gateway reconfiguration). Additionally, deploy a regex complexity budget or execution timeout at the application or reverse-proxy layer to abort runaway evaluations before they saturate CPU; this mitigates the impact if a malicious pattern is submitted but does not fix the root cause. Monitor worker process CPU usage as a detection signal for active exploitation attempts.
Same technique Denial Of Service
View allVendor StatusVendor
Share
External POC / Exploit Code
Leaving vuln.today