VulcanSphere (generated with Stable Diffusion 3.5 Large) · CC0 1.0 Universal Public Domain Dedication; in jurisdictions where the work is not copyrightable (including the United States) it is public domain as an AI-generated work without sufficient human authorship. Commercial reuse permitted. Provenance and license verified on Wikimedia Commons on October 6, 2026.
ANALYSIS AI Safety Research
Keep It CALM: A Geometric Case That One Global 'Unsafe Direction' Cannot Fix Text-to-Image Safety Everywhere
A NeurIPS 2026 paper argues that the dominant design of training-free image safety filters, which removes one reusable unsafe subspace across all prompts, faces a structural coverage-selectivity trade-off. Its proposed fix, CALM, routes each prompt locally. This analysis explains the argument for general readers, and flags which parts are demonstrated versus self-reported.