OpenAI Provider EUVD-2026-43057

Exploitation requires an authenticated session with Drupal administrator-level privileges (PR:H per CVSS vector), meaning the attacker must either already hold admin credentials or have compromised an admin account. High attack complexity (AC:H) further constrains exploitation, implying the attacker must know or enumerate valid internal network targets. The Drupal OpenAI Provider module must be installed and active on the target site - it is a contributed module, not part of Drupal core, and must be explicitly enabled. The SSRF does not operate against default, unconfigured Drupal installations. External unauthenticated attackers cannot trigger this vulnerability. Exploitation is limited to scenarios involving insider threat, compromised admin credentials, or privilege escalation chains.

All available risk signals align at low severity. The CVSS 3.3 base score reflects three compounding mitigations: high attack complexity (AC:H), high privilege requirement (PR:H), and limited confidentiality and integrity impacts (C:L/I:L/A:N). An attacker must already hold Drupal administrator credentials - an extremely powerful role - making external exploitation implausible outside a compromised-admin or insider scenario. EPSS at 0.14% (4th percentile) confirms negligible exploitation likelihood in the wild. SSVC corroborates this: Exploitation none, Automatable no, Technical Impact partial. The non-default, niche nature of a contributed AI-integration module further constrains the population of reachable targets. This vulnerability does not warrant emergency patching but should be addressed in normal patch cycles, particularly for sites where Drupal admin credentials are shared among multiple users.

An authenticated Drupal administrator navigates to the OpenAI Provider module configuration and submits a crafted URL - for example, an internal cloud metadata endpoint or an intranet service address - as a parameter the module passes to its HTTP client. The Drupal server fetches the target URL server-side, potentially returning internal response data visible in the admin UI or logs. No public exploit code exists; the attacker must already possess admin credentials and knowledge of internal network topology.