OpenColorIO EUVD-2026-38769

Exploitation requires the victim to load a crafted .spi3d LUT file through OpenColorIO's file parser on a non-Windows platform (the description states the overflow occurs on non-Windows; Windows is noted as not affected by this overflow). The specific trigger is a LUT data line whose token exceeds the 64-byte destination buffer, read from the 4096-byte lineBuffer via sscanf %s at FileFormatSpi3D.cpp:163. Limiting factors: the attacker needs no authentication (PR:N) but does need user/pipeline interaction to open or process the malicious file (UI:A), the affected version must be earlier than 2.5.2, and the attack is local/file-based (AV:L) rather than remotely network-triggerable.

The vendor CVSS 4.0 vector (AV:L/AC:L/AT:N/PR:N/UI:A/VC:H/VI:H/VA:H, base 8.4) characterizes this as a local, low-complexity, unauthenticated issue that requires victim interaction (UI:A) - i.e., the attacker does not need an account, but a user or pipeline must open/process the malicious .spi3d file, which is the dominant limiting factor. There is no CISA KEV listing and no public exploit identified at time of analysis, and no EPSS score was provided, so there is no signal of active or imminent mass exploitation. The realistic risk is therefore moderate: high technical impact (full C/I/A on the local process) and trivial complexity, but gated by the need to deliver and have the victim load a crafted file, making it better suited to targeted supply-chain or asset-poisoning attacks (e.g., a malicious LUT distributed through a VFX asset pipeline) than to opportunistic remote attack. Note the framing as 'local' (AV:L) is appropriate because the input is file-based rather than network-delivered, even though the file could originate remotely.

An attacker crafts a malicious .spi3d LUT file with an overly long data token and distributes it through a shared VFX asset library, project deliverable, or downloaded look pack. When an artist or an automated render pipeline opens the file in an OCIO-backed application on Linux or macOS, the oversized token overflows a 64-byte stack buffer by roughly 4000 bytes, corrupting the stack and potentially leading to crash or code execution in the host process. No public exploit code has been identified, and successful exploitation depends on the victim choosing to load the file (UI:A).