Policy · Ethics · Privacy

“Is this allowed?” Research AI Guardrails Assistant

A chat-style guide that quickly checks if your research AI workflow is Green, Yellow, or Red—and offers safer alternatives with approved tools.

Local-first. No analytics. Answers stay on your device; evaluation runs in-app.

Sample outputYellow

Human data, IRB pending

  • Stop external uploads until IRB is approved.
  • Use local/offline tools or synthetic data.
  • Document protocol and de-identification steps.
Checklist: status YELLOW. Destination not approved. Use local tools; log rationale in IRB notes.

Deterministic, local-first evaluator (no external calls)

3–6 guided questions with policy-aware routing

Instant Green / Yellow / Red plus safer workflow

Copy-ready checklist for email/IRB notes

Browsable rules + examples for transparency