AI incident runbooks (ten, free)
First fifteen minutes, first hour, evidence, containment, durable controls, and closure for ten AI incident types, from leaked model keys to coding-agent CI compromise.
Runbooks you can run without us, the document we sign before adversarial work, what we publish in the open, and the descriptions partners and press can use as written.
First fifteen minutes, first hour, evidence, containment, durable controls, and closure for ten AI incident types, from leaked model keys to coding-agent CI compromise.
The document signed before any adversarial work: targets, identities, permitted and prohibited actions, stop conditions, evidence handling.
What Qompute AI publishes in the open, on Hugging Face and here, with what each item does and does not show.
Short descriptions of Qompute AI at 25, 50, and 100 words, plus two introduction notes a partner can send as written.
Mark and wordmark with clear-space and misuse guidance.
A one-page executive summary for qualified sharing.