AMD Compressonator SDK CVE-2026-40661

Exploitation requires that a user or an automated pipeline running the AMD Compressonator SDK process an attacker-supplied, specially crafted DDS (DirectDraw Surface) file. Limiting factors: the attack vector is local (AV:L) with no remote/network exposure, and it depends on user interaction (UI:P) - the malicious DDS file must actually be opened or ingested by the Compressonator-based tool or pipeline. There is no remote, unauthenticated, or default-service exposure; the flaw only manifests when the SDK is invoked to parse untrusted DDS input.

This is a medium-severity, locally-triggered issue that is not an urgent priority. The CVSS 4.0 vector (AV:L/AC:L/AT:N/PR:L/UI:P) confirms exploitation requires local delivery of a malicious file plus user interaction, not remote network exploitation. The impact profile is skewed toward availability (VA:H) with only limited integrity impact (VI:L) and no confidentiality impact (VC:N) - consistent with the described heap corruption / denial-of-service outcome rather than confirmed code execution. There is no public exploit identified at time of analysis: the input provides no POC reference, no EPSS score, and no CISA KEV listing, so there is no evidence of active exploitation. The vuln.today decision label is 'this_month,' which aligns with a real-but-moderate issue rather than act_now/emergency. The CWE-787 out-of-bounds write class can in principle escalate beyond DoS to code execution, but the available evidence states only 'heap corruption or denial of service (DoS)' - code execution is not claimed, so it should not be assumed. No signals conflict here; the main gap is the absence of EPSS/exploit-maturity data, which is consistent with a developer-tooling flaw unlikely to see broad opportunistic exploitation.

An attacker crafts a malicious DirectDraw Surface (DDS) texture file and delivers it to a developer or an automated asset-processing pipeline that uses the AMD Compressonator SDK. When the victim or pipeline processes the file, the out-of-bounds write triggers heap corruption, crashing the process or disrupting the build/conversion pipeline. No public exploit code is identified at time of analysis.