Analysis
Prompt Injection Vulnerability Disclosure Framework
First reported · Discovered permsai.com
Page published
Earliest dated coverage: 13 Sep 2026 · First observed: 12 Oct 2026 · Latest dated coverage: 13 Sep 2026
Coverage timeline
Single-source analysis — one report is available.
Why it matters
The PermsAI framework helps defenders avoid turning a persuasive prompt-injection transcript into an unsupported vulnerability claim, improving triage and disclosure quality across AI deployments.
PermsAI publishes an editorial framework for classifying prompt-injection vulnerability disclosures, distinguishing model behavior, product weakness, application design flaw, security-boundary failure, software vulnerability, and actionable disclosure. The guidance walks security teams through stating the intended security property, identifying the trust boundary, classifying the input (direct/indirect/retrieved/tool/memory), reproducing with controlled context, and assigning claim strength before coordinated disclosure.