AWS details Amazon Quick production security: shape the data, don't rely on permissions alone
AWS ML Blog · rss · 2026-09-02
Amazon Quick BI projects often pass a small pilot, then stall when security and compliance review the production plan — a permission model for ten pilot users breaks across five departments. Using the AnyCompany scenario (5,000 employees, 5 departments, 5 locations), AWS shows how to make security hold at scale.
Four validated patterns (shared idea: remove data before it reaches users, rather than blocking via permissions)
- Dataset shaping: one source dataset becomes three — full 30 columns for HR leadership; a manager version with 4 sensitive columns removed plus Row-Level Security; an all-employee version with only 25 aggregated rows. No permission mistake can expose a column that doesn't exist.
- Agent isolation: each Agent connects to exactly one dataset scoped to its audience.
- Document classification: sensitive documents are excluded from the knowledge base entirely.
- Approval gates: all outbound actions in Flows require human-in-the-loop review.
The post includes a nine-step walkthrough (RLS rules CSV, permission datasets, CloudTrail auditing), a governance framework, and a production readiness checklist.
More from Safety
- Anthropic Releases Claude Fable 5.1 and Mythos 5.1: Lower Costs and Enhanced Privacy — TFenrir · 2026-09-02
- Anthropic Releases Claude Fable 5.1 and Mythos 5.1: Lower Costs and Enhanced Privacy — scaling01 · 2026-09-02
- Anthropic releases Claude 5.1: 25% cheaper with zero-retention enterprise safeguards — claudeai · 2026-09-02
- Doberman: An execution gateway to prevent agent disasters — Da_Lil_Fu · 2026-09-02
- Harvard Dean criticized for using AI-generated content — soumitrashukla9 · 2026-09-02
- ECLIPSE: Self-Evolving Stealthy Attack on Long-Horizon Agents — chaumian · 2026-09-02