New AI security tool lays out key exposures 

The article introduces the Cloud Security Alliance’s Risk Rubric, an AI risk leaderboard that assigns A–F grades to LLMs across six pillars (security, reliability, privacy, safety, transparency, reputation) using automated testing, documentation reviews, and adversarial red teaming. Experts explain how organizations can use these scores to support model selection, procurement thresholds, and ongoing governance, while acknowledging challenges such as model drift, transparency, and potential gaming; the piece positions Risk Rubric as one part of a broader AI risk strategy alongside ML bills of materials and software supply chain hardening.