Validate · Assess & Decide · Writing & Content
Privacy-by-design seven-principles AI register
A JSON register with one entry per PbD principle passes validation: all seven canonical principles are present with correct slugs, each has an implementation note of minimum length, a status from the allowed enum, an assigned owner, and no principle is duplicated.
You receive: A JSON object with 'system_name' (string), 'assessment_date' (ISO date), and 'principles' (array of 7 objects). Each principle: principle_slug (enum: proactive-not-reactive|privacy-as-default|privacy-embedded-in-design|full-functionality|end-to-end-security|visibility-and-transparency|user-centric), implementation_note (string >= 50 chars), status (enum: implemented|partial|not-started), owner (string non-empty).
Part of Protect IP & Data
Example
A sample of what this play produces. Your result is generated for your inputs.
- System name
- STUD.com play platform (workspace facts, play runs, acceptance judge)
- Assessment date
- 2026-08-17
| Principle slug | Implementation note | Status | Owner |
|---|---|---|---|
| proactive-not-reactive | Every play ships with an acceptance check authored before the play goes live, so data-handling defects are screened at design time, before a buyer's run, not patched after an incident. The gate blocks a deploy when a play's contract drifts from its validator. | implemented | Dan Schmitz, founder |
| privacy-as-default | A workspace's facts are private to that workspace by default: intakes prefill only from the buyer's own workspace, no fact is shared across accounts, and paid fulfillment stays gated off until auth and payment rails are reviewed. | implemented | Dan Schmitz, founder |
| privacy-embedded-in-design | Workspace facts are a first-class schema (key, value, provenance) rather than free text pasted into prompts, so what an agent sees per run is enumerable, and the judge harness runs process-isolated from buyer data stores. | implemented | Dan Schmitz, founder |
| full-functionality | The living-system value (state a fact once, every run uses it) is delivered without asking buyers to trade privacy for utility: prefill works from the buyer's own workspace only, and no cross-buyer data pooling is required for any play to run. | implemented | Dan Schmitz, founder |
| end-to-end-security | Credentials live in the deploy platform's build environment, never in CI; the database is a managed Postgres with TLS; deletion and retention procedures for workspace facts are drafted but not yet exercised end to end, so this stays partial. | partial | Dan Schmitz, founder |
| visibility-and-transparency | Each play publishes its honest scope, its intake schema, and the named checks its acceptance judge runs, so a buyer can read exactly what the system does with an input before commissioning a run. A public deliverable-kind gallery shows output shapes. | implemented | Dan Schmitz, founder |
| user-centric | Buyers own their workspace facts and edit or remove them directly in the workspace UI; intakes show what will be used before a run starts. A self-serve export of all facts and deliverables is on the roadmap but not shipped, so this stays partial. | partial | Dan Schmitz, founder |
Opens soon
STUD opens runs for everyone soon, on every play at once.
Get early access to STUD the day it goes live.