NIST AI RMF — Self-Alignment
The NIST AI Risk Management Framework (AI RMF 1.0) is a voluntary framework for managing the risks of AI systems, organized into four functions — Govern, Map, Measure, Manage. This page maps OWL's optional AI features — the QUAIL assistant and AI-assisted modeling — to those functions. It pairs with our EU AI Act position and our "no training on your data" commitment.
Voluntary framework, no certification. NIST does not certify conformance to the AI RMF — it is a voluntary tool for organizing AI-risk practices. This page is an honest self-alignment, not a certification, and every claim matches our privacy policy.
1. Mapping OWL's AI to the AI RMF functions
| AI RMF function | How OWL aligns |
|---|---|
| Govern | AI in OWL is optional and opt-in — if you never use it, nothing is sent to any model. Privacy guardrails are built in: no account identity, files, or email accompany a prompt; we keep no copy of prompts, images, or responses beyond serving the request; and a no-training guarantee applies to us and is required of any third-party provider. The model may be self-hosted or a third-party provider (subject to change), disclosed on the sub-processor page. |
| Map | We document where AI is used and what is sent: QUAIL relays your prompt text; AI-assisted modeling can additionally send rendered images of your scene (or reference images you provide) so the model can "see" your work. Intended use is productivity assistance for modeling and chat. Foreseeable misuse is bounded by the data limits above and by the fact that image features transmit pictures of your workspace — which we tell users plainly so they can choose not to use them. |
| Measure | We record per-request usage metrics — model name, token counts, and status — tied to your account for rate-limiting and billing, which also gives us aggregate reliability signals. AI features are clearly labeled as AI so users can judge output accordingly. We do not run hidden background AI on your content. |
| Manage | Users control whether AI runs at all (opt-in, per feature). The no-training requirement is a contractual control on providers; we retain the ability to change or withdraw the AI provider as the service evolves. Issues or concerns can be raised through support and we adjust the features and disclosures accordingly. |
2. Scope & honesty
This is a voluntary, point-in-time self-alignment of OWL's AI features with the NIST AI RMF — not an audit or certification (the framework offers none). It is consistent with the privacy policy and our EU AI Act position. If you identify an AI-risk concern, please reach us via the disclosure / contact channels.