FFmpeg developers publicly call out AMD over a suspected AI-generated patch, reinforcing strict human review standards to protect critical open source infrastructure.
Open source governance took centre stage after FFmpeg maintainers accused AMD of submitting unchecked, AI-generated code to the widely used multimedia project, publicly criticising the contribution as “AI slop” and rejecting the patch.
AMD had proposed changes to FFmpeg, one of the longest-running open-source video and audio processing tools that underpins Chrome, Handbrake, OBS Studio and VLC Media Player. Developers reviewing the submission said the code appeared AI-generated and insufficiently vetted by a human.
Zhao Zilli, an FFmpeg developer, wrote: “please ensure a thorough manual review of all AI-generated code before submission.” Zilli added, “AI-generated code tends to be verbose; it requires the developer’s own judgment and cleanup,”
The project’s official X account escalated the issue, posting: “Hi @AMD. FFmpeg developer @quink_lamy is not happy with your AI slop patches.”
Maintainers flagged unnecessary extra comments, verbose structures typical of AI output, non-standard coding techniques, weak optimisation and signs of missing manual review.
The AMD developer who submitted the patch denied using AI, stating: “From my practice, those steps […] are necessary. This is not related AI or human. This is really my experience and my practices.” The developer said the comments were intentional and offered to remove them.
The patch aimed to add support for AMD’s HIP SDK on Windows to enable GPU-accelerated video processing, but was rejected on technical grounds, with maintainers noting similar capability already exists through Vulkan.
