{"id":"CVE-2026-117574","published":"2026-10-28T20:17:21.417","lastModified":"2026-10-28T20:17:21.540","description":"A flaw was found in ai-gateway-payload-processing. An authenticated remote attacker can exploit this vulnerability by submitting a request specifying another tenant's model identifier in the request body or header. Because the service does not verify authorization before retrieving credentials from the shared store, it injects the target tenant's application programming interface (API) keys into the request. This flaw allows an attacker to perform unauthorized requests and consume another tenant's large language model (LLM) resources and quotas.","cvssScore":9.1,"cvssSeverity":"CRITICAL","cvssVector":"CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:C/C:H/I:L/A:L","cwes":["CWE-639"],"vendors":[],"products":[],"references":[{"url":"https://access.redhat.com/security/cve/CVE-2026-117574","tags":[]},{"url":"https://bugzilla.redhat.com/show_bug.cgi?id=2494861","tags":[]}],"exploitRefs":[],"hasPoc":false,"ai":{"summary":"An authenticated attacker can abuse insufficient authorization checks in ai-gateway-payload-processing to inject another tenant's API keys into their own requests by specifying the target tenant's model identifier, enabling unauthorized consumption of the victim's LLM resources.","exploitability":"Requires valid authentication and knowledge of a target tenant's model identifier; low effort once those conditions are met.","blast_radius":"An attacker can exhaust another tenant's API quota, incur unexpected costs, and potentially disrupt their AI workloads by consuming their allocated LLM capacity.","remediation":"Upgrade to ai-gateway-payload-processing version 2.590 or later, where the service now enforces tenant-scoped authorization before retrieving model credentials from the shared store.","detection":"Look for requests to the payload-processing endpoint where the authenticated user's tenant ID does not match the tenant ID embedded in the model identifier used in the request body or header; such mismatches are strong indicators of unauthorized cross-tenant access attempts.","tags":["authorization-bypass","multi-tenancy","resource-abuse","api-key-injection","llm"],"model":"garnet-coder","analyzedAt":"2026-10-29T08:45:51.074Z"}}