EnergizeLab AI Station CVE-2026-105036

Requires the attacker to occupy a man-in-the-middle position on the network path between the affected device/app and its update server (e.g., ARP spoofing on the shared LAN, a rogue Wi-Fi access point, or a compromised router/upstream link), because updates are fetched over cleartext HTTP with no signature or hash verification. Exploitation is also gated by an update check actually occurring (attacker may need to wait for or induce an update), and reachability depends on whether update traffic stays on the LAN (adjacent, AV:A) or crosses the internet. No authentication and no user interaction are required once the MITM position is held, but the privileged-network-position prerequisite is the primary limiting factor and makes default-configuration remote exploitation from an arbitrary internet host infeasible.

The provided CVSS 3.1 vector (AV:A/AC:H/PR:N/UI:N/S:U/C:H/I:H/A:H, base 7.5) describes an unauthenticated but adjacent-network, high-complexity attack yielding full confidentiality, integrity and availability impact - consistent with CWE-494 (download of code without integrity check), where a man-in-the-middle substitutes a malicious firmware update delivered over plain HTTP. The AC:H and AV:A metrics are the key risk dampeners: the attacker must already hold a privileged position on the network path (LAN MITM, rogue AP, or compromised router/upstream) and win against the update transaction, so this is not a spray-and-pray internet-wide bug. No EPSS score, no CISA KEV listing, and no vendor-confirmed exploitation were supplied, so there is no signal of active or widespread exploitation. POC availability is unclear: one reference is a GitHub repository ('your-secrets-are-safe-with-me') whose exploit status is unverified, so it is NOT counted as confirmed public exploit code. Note a vector tension worth verifying with the vendor: MITM against an update channel that traverses the public internet could arguably justify AV:N rather than AV:A - if the devices fetch updates over WAN rather than only LAN, real-world reachability is broader than AV:A implies. Overall this is a genuine but conditional/high-barrier risk, not a high-CVSS-low-real-risk paper cut, because successful exploitation gives full device takeover via attacker-controlled firmware.

An attacker positioned on the same network segment as an Eilik/EnergizeLab device (or on a compromised router/AP in the update path) intercepts the plaintext HTTP firmware-update request and returns a crafted malicious firmware image, which the device or Windows app installs without checking any signature or hash. The attacker thereby gains full control of the robot/station or the updating host. No confirmed public exploit is identified at time of analysis, though one referenced GitHub repository may contain related material.