Chamilo LMS CVE-2026-114389

Requires a valid authenticated Chamilo account of any role (e.g., an ordinary student), per CVSS PR:L - not remote-unauthenticated. It also requires that the AI grading feature backing the open_answer_grade endpoint (AiController::openAnswerGrade) is configured/enabled on the instance, and the attacker must enumerate or guess valid exeId and questionId values corresponding to other students' exam attempts. Limiting factors that reduce real-world risk: authentication is mandatory, the vulnerable AI-grading feature must be active, impact is confined to disclosure of other students' open-answer content via feedback plus AI-quota consumption (no write/integrity or service-availability impact), and no user interaction from the victim is needed.

This is a medium-severity authorization flaw (CWE-639, IDOR) whose signals are internally consistent and point to a real but bounded issue rather than an emergency. The provided CVSS 4.0 vector (AV:N/AC:L/AT:N/PR:L/UI:N/VC:L/VI:N/VA:N) scores 5.3: network-reachable and low-complexity, but PR:L means an attacker must already hold a valid authenticated account, and impact is limited to low confidentiality (VC:L) with no integrity or availability impact. No EPSS score was provided, there is no CISA KEV listing, and no public POC is referenced - so by the precision rules this is 'no public exploit identified at time of analysis.' The main exploitation primitive is trivial parameter enumeration (exeId/questionId), which lowers the technical barrier, but the authentication requirement and the narrow impact (viewing peers' open-answer content via AI feedback and burning AI quota) keep real-world urgency moderate. Net: a legitimate patch-worthy issue - especially for institutions where many low-trust student accounts exist - but not a high-priority internet-wide threat. No signals are in direct conflict; the primary gap is the absence of EPSS/POC data, which limits confidence about exploitation likelihood in the wild.

A logged-in student on a Chamilo LMS instance (2.0.0-alpha.4 through pre-2.0.0-RC.3) sends crafted requests to the open_answer_grade endpoint while incrementing exeId and questionId values belonging to other students' exam attempts. Because AiController::openAnswerGrade never verifies the caller's authorization over the target course/attempt, the server returns AI-generated grading feedback that discloses other students' answers and consumes the institution's AI quota. No public exploit code is identified, but the attack requires only basic HTTP request manipulation given valid credentials.