The image format is the smaller news here. What I notice is how it shipped: Chrome did not bolt on the C++ reference decoder, it integrated a Rust reimplementation, because image decoders chew untrusted bytes straight off the network. Anyone self-hosting a service that parses uploaded files has the same problem at smaller scale. The sandbox is the second layer, not the first.
What I’d do: nothing drastic. Chrome’s own advice is to try both AVIF and JPEG XL, with JXL expected to win on high-fidelity or lossless photos. So I’d encode a few photos both ways and compare. Lossless JPEG transcoding is the feature I’d test first.
The story — Chrome supports decoding JPEG XL (.jxl) from version 155. The format offers 30–50% better compression than JPEG, lossless compression, built-in HDR and lossless JPEG transcoding. The decoder is jxl-rs, a Rust implementation using a SIMD abstraction layer that confines unsafe operations to a few reviewed locations. The team verified it with fuzzing and AI code review and found no memory safety bugs. Developer requests through Interop drove the decision. (Source)