FreeRDP CVE-2026-109253

Exploitation requires a reverse-RDP setup: the FreeRDP-based client must be using Remote Credential Guard AND must connect to a malicious or compromised RDP server, which then sends a crafted unaligned RDPEAR payload (e.g., a CreateApReqAuthenticatorReq message) to trigger the out-of-bounds read in the NDR scalar readers. Limiting factors that reduce realistic risk: the attacker must control or compromise the server the victim connects to (not a push against a listening client), the victim must actively initiate the connection (UI:P), Remote Credential Guard must be in use (a non-default, specific deployment mode), crafting a valid unaligned payload raises complexity (AC:H/AT:P), and only FreeRDP versions 3.9.0 through 3.31.x are affected (fixed in 3.32.0).

This is a moderate-risk, low-urgency issue - a reverse-RDP denial-of-service rather than a priority emergency. The CVSS 4.0 signals are internally consistent and point to limited real-world risk: AV:N but AC:H (crafted unaligned payload required), AT:P (a specific attack precondition must be present), UI:P (the victim must initiate a connection to the attacker's server), and impact restricted to availability (VA:H) with no confidentiality or integrity effect (VC:N/VI:N). PR:N reflects that the threat actor is the RDP SERVER and needs no credentials on the client - not that a classic unauthenticated remote attacker can reach a listening service. There is NO EPSS score, NO SSVC decision, and NO KEV entry provided, and no public exploit is indicated, so there is no public exploit identified at time of analysis. Note a conflict worth flagging: the input Tags include 'Information Disclosure,' but the official CVSS 4.0 vector sets VC:N (no confidentiality impact) and the description frames the realized consequence as a crash (over-read → DoS); treat the information-disclosure framing as the CWE-125 vulnerability class, not a confirmed data-leak impact - verify with the FreeRDP advisory if leak potential matters.

A user running a FreeRDP-based client with Remote Credential Guard enabled is lured or redirected into connecting to a malicious or already-compromised RDP server. During the RDPEAR exchange the server returns a crafted unaligned payload such as a CreateApReqAuthenticatorReq message; the client's NDR scalar reader checks value length before consuming alignment padding and reads out of bounds, crashing the client process. No public exploit is identified at time of analysis, and exploitation is high-complexity.