An insecure direct object reference allows any authenticated user to retrieve sensitive metadata about AI personas, features, and LLM models by specifying their identifiers. The data exposed includes credit allocations and usage statistics that are not intended for public consumption. The vulnerability results in a modest confidentiality impact and does not affect integrity or availability.

Discourse open‑source discussion platform is affected. Versions earlier than 2026.3.0‑latest.1, 2026.2.1, and 2026.1.2 are vulnerable. All releases identified in the advisory but not yet patched are susceptible.

The CVSS score of 5.3 indicates moderate risk, and the EPSS score of less than 1% suggests a low likelihood of current exploitation. The vulnerability is accessed over the network by any logged‑in user, making it easy to exploit without special privileges. It is not listed in the CISA KEV catalog. While the impact is moderate, a malicious actor could gather usage data that may be valuable for profiling or exploiting the platform.