REBUILD EUVD-2026-101239

Requires a valid authenticated low-privilege account on the target REBUILD instance (through 4.5.0) with the AiBot2 assistant feature in use. The attacker must supply another user's chatid to the /aibot2 chat-init, chat-delete, chat-feedback, or chat-stream-stop endpoints, which act on the caller-supplied chatid without verifying ownership. The key limiting factor is obtaining a valid victim chatid - reflected in AC:H - so predictability or enumerability of chatids directly governs exploitability; exploitation is constrained to authenticated users and has no effect on single-user or non-multi-tenant deployments. No user interaction by the victim is required.

This is a moderate-severity authorization bypass (IDOR, CWE-639) whose real-world urgency is tempered by its access requirements. The CVSS 4.0 vector (AV:N/AC:H/AT:N/PR:L/UI:N) confirms network reachability but requires an authenticated low-privilege account (PR:L) and carries high attack complexity (AC:H), reflecting the need to obtain or enumerate another user's chatid rather than exploit a default/predictable value. Impact is skewed toward confidentiality (VC:H - reading/exfiltrating other users' AI conversations) with only limited integrity impact (VI:L - deleting or interrupting conversations) and no availability or scope impact. Exploitation status: no public exploit identified at time of analysis, though a detailed technical write-up (hackmd) with direct vulnerable source-code references exists and materially lowers the barrier to developing one. EPSS is not provided in the input, so exploitation probability cannot be quantified. KEV status is NO - no confirmed active exploitation. On balance this is a legitimate but not emergency-tier issue: it affects authenticated users only, impacts a niche open-source product, and lacks evidence of weaponization. It merits prompt patching for exposed multi-tenant REBUILD deployments but does not rank with unauthenticated RCE or KEV-listed flaws.

A low-privilege but authenticated user of a shared REBUILD instance obtains or enumerates another user's chatid and submits it to an /aibot2 endpoint (e.g. chat-init or chat-stream-stop) to read, exfiltrate, delete, or interrupt that victim's AI assistant conversation. No public exploit code is identified, but a published technical write-up pinpoints the vulnerable controller and ChatManager code paths, making exploitation straightforward for anyone with an account.