Lunary

Lunary provides a focused artificial-intelligence and language-model platform, and despite its narrow product scope, occupies a prominent position in the vulnerability landscape for emerging AI infrastructure. Vulnerabilities affecting this vendor skew strongly toward critical-severity outcomes, concentrating around authorization and access-control weaknesses—including missing authorization checks, authorization bypasses through user-controlled keys, and improper authorization logic—that are characteristic of systems managing sensitive model access and data flows. The recurring pattern reflects the access-control complexity inherent to multi-tenant AI platforms where misconfigured permissions can expose models, training data, or inference results across user boundaries. Defenders should treat Lunary disclosures as high-priority, particularly in environments where the platform gates access to proprietary or sensitive language models. Current exploitation activity, severity breakdowns, and exposure counts are shown alongside this summary.