ByteBrief
We're a portrait publication through and through. Turn your phone back and your briefing picks up right where you left it.
(We tried widescreen once. It wasn't us.)

A 32-bit integer truncation in vLLM's GGUF dequantization kernels causes output tensors to be partially filled with uninitialized GPU memory. In multi-tenant serving, that memory can contain another user's data. The vulnerability is rated CVSS 5.3 and affects four functions in gguf_kernel.cu that use torch::empty followed by a kernel bounded by a narrowed dimension.
Tap to vote and see what everyone thinks.
Summary by ByteBrief