Autodesk Fusion EUVD-2026-75449

A maliciously crafted Autodesk Fusion add-in must be installed and executed by the target user on Fusion Desktop versions 2705.0.0 through 2705.1.10 - user interaction to install and run the add-in (UI:R) is the primary limiting factor. The attacker must also operate or control a network-reachable proxy server to receive and inspect the redirected traffic. No elevated OS privileges are required by the add-in itself; the Fusion application context provides sufficient access to overwrite proxy settings. The vulnerability does not affect Fusion 2705.1.11 or later. There is no remote, unauthenticated path - exploitation is constrained to scenarios where the user actively installs and runs the malicious add-in.

The vendor-assigned CVSS 8.6 (AV:L/AC:L/PR:N/UI:R/S:C/C:H/I:H/A:H) captures the scope change - proxy settings affect traffic routing beyond the immediate application process boundary - and reflects a worst-case CIA triple. However, the description's stated consequence is primarily information disclosure ('potentially exposing sensitive information'), making the I:H and A:H components appear overstated relative to the described impact; while a redirecting proxy could theoretically alter or block traffic, neither outcome is described as a confirmed effect of this vulnerability. Real-world risk is gated by the social engineering requirement of convincing a professional CAD user to install a malicious add-in (UI:R), which is the primary limiting factor. No EPSS data, KEV listing, or public POC code was included in the input, limiting assessment of opportunistic exploitation risk. The attack is plausible in targeted scenarios such as supply-chain compromise of a legitimate community add-in or distribution of a trojanized utility through engineering forums.

An attacker publishes a free Autodesk Fusion add-in - for example, a parametric modeling utility - to a community forum or third-party marketplace; an engineer installs and runs it during normal design work. The add-in silently writes an attacker-controlled proxy server address to Fusion's persistent proxy configuration; from that point forward, every authenticated Fusion cloud API call is routed through the attacker's proxy, allowing interception of session tokens and potentially proprietary CAD data without any further user interaction. No public proof-of-concept code has been identified at time of analysis.