UltraVNC CVE-2026-7830

Exploitation requires the target connection to use the legacy MS-Logon II authentication scheme (rfbUltraVNC_MsLogonIIAuth); connections using MS-Logon III (X25519 + AES-256-GCM) are immune. The attacker must be able to observe the authentication handshake on the network - via passive sniffing on a shared/switched segment, traffic recording, or an active man-in-the-middle/on-path position - which is the primary limiting factor and the reason CVSS scores AC:H rather than AC:L. No authentication is required (PR:N) and no user interaction beyond the victim initiating a normal login (UI:N). The attack is confined to UltraVNC on Windows through 1.8.2.2 and only yields credentials from handshakes the attacker can actually capture, so an attacker with no network vantage point cannot exploit it remotely at scale.

This is a genuine credential-exposure risk for environments still using legacy MS-Logon II, but exploitability is gated by network position rather than being drive-by remote. The CVSS 3.1 vector (AV:N/AC:H/PR:N/UI:N/S:U/C:H/I:H/A:N, 7.4) correctly encodes that no authentication or user interaction is needed but that the attacker must be able to observe the handshake - hence AC:H. No EPSS score, KEV entry, or POC was provided, so there is no evidence of active exploitation; the cryptanalysis is well understood and reproducible, but weaponization requires privileged traffic access (LAN sniffing, ARP spoofing, or an on-path MITM). Net: prioritize where MS-Logon II is in use on untrusted or shared networks, but this is not an internet-scannable, mass-exploitable bug. The I:H rating in the supplied vector is debatable - the direct primary impact is confidentiality (credential disclosure); integrity impact is a downstream consequence of using the recovered credentials, so treat the I rating with caution and see the assessed vector below.

An attacker on the same LAN segment as a VNC client and server (or positioned on-path via ARP spoofing) passively records the MS-Logon II handshake. They solve the 64-bit Diffie-Hellman exchange with Pollard's rho and recover the time-seeded private exponent by brute force within a minute, reconstruct the shared key, and decrypt the encapsulated username and password to obtain the operator's credentials. No public POC was provided, but the technique is fully described and reproducible from the CVE detail, and the AC:H rating reflects the required observation position.