CVE Alert: CVE-2026-125201 – FFmpeg

FFmpeg 7.0 before 8.1.3 and 9.x before 9.0.2 contains a heap out-of-bounds write vulnerability in vk_av1_decode_slice() in libavcodec/vulkan_av1.c because the tile-count limit is checked only on entry. Attackers can supply a crafted AV1 stream with more than 256 tiles decoded via the Vulkan hardware accelerator to overflow tile_sizes and corrupt heap memory.

**Risk verdict:** High risk where untrusted AV1 media is processed with Vulkan acceleration; no supplied evidence confirms active exploitation, so urgency depends on exposure and whether that decode path is enabled.

**Why this matters:** Successful heap corruption could compromise the media-processing process, affecting confidentiality, integrity and availability. The most realistic immediate outcome is a crash or disruption; further compromise is possible but not established by the supplied data. KEV, SSVC, PoC and EPSS indicators were not provided, leaving exploitation likelihood uncertain.

**Most likely attack path:** An attacker must get a crafted AV1 stream processed locally, with the vulnerable Vulkan decode path active; this is not a direct remote attack. No prior privileges are indicated, but user interaction and a present attack condition are required, and exploitation is assessed as difficult. Scope is unchanged, so impact is expected within the affected process rather than automatically extending to other systems.

**Who is most exposed:** Media servers, transcoders, desktop applications and pipelines that ingest untrusted video and enable Vulkan hardware decoding are the main concern.

Identify workloads invoking Vulkan-backed AV1 decoding.

Alert on decoder failures or abrupt worker exits while processing AV1.

Correlate heap-corruption or allocator diagnostics with media-ingestion events.

Review crash dumps for faults in AV1 Vulkan decoding.

Mitigation and prioritisation:

Upgrade to a fixed upstream release; prioritise exposed media-processing systems.

Until patched, disable Vulkan AV1 decoding or force a safer software-decoding path.

Restrict and validate untrusted media before it reaches shared processing services.

Test decoder changes against representative workloads and stage rollout to avoid service disruption.

A considerable amount of time and effort goes into maintaining this website, creating backend automation and creating new features and content for you to make actionable intelligence decisions. Everyone that supports the site helps enable new functionality.

If you like the site, please support us on Patreon or Buy Me A Coffee using the buttons below.