Profi · Topic
Ethics as ops gate
Quality, IP, training, and privacy checks are part of shipping — not a slide after launch.
Purpose
CHI designers flagged ethics and literacy; vendor FAQs and USCO materials make IP/training concrete. Profi track treats these as an operating gate on every handoff: if we cannot answer training, ownership, and privacy, we do not ship the generation.
Notes
- **Training & data** — know whether customer files train the vendor model; opt out when policy requires (`figma-ai-terms` / FAQ).
- **IP & registration** — AI-assisted work may need human authorship documentation (USCO study/hub).
- **Quality ≠ polish** — Designlab 2026: designers distrust “pretty but wrong”; keep `human-method-gate` on the critical path.
- **Responsible vendor claims** — Adobe-style responsible-AI pages are resources; practice still needs a named owner for each ship.
- **Product AI vs process AI** — designing AI features is a different job (`nng-four-meanings-ai-design`); do not smuggle it into “we used Make.”
When to use / Skip
Use before any external publish, client delivery, or model fine-tune on company data. Skip when the work never leaves a private scratch file and will be fully redrawn by hand — still apply `human-method-gate`. Falsifier: “the vendor said it’s fine” without reading the terms for your plan tier.