Intended-use cases

An assessment is always bound to an intended use. Each use below maps to a distinct right in the license library.

• Commercial use of licensed digital assets — May an agent acquire and commercially exploit this asset? a2ally checks the controlling terms' reviewed profile for the commercial_use permission and any prohibition before any purchase.
• Using assets for AI model training — Training use is a distinct right from inference. a2ally assesses model_training separately: prohibited training returns BLOCK, undocumented training INSUFFICIENT_EVIDENCE, and training under licenses whose conditions may or may not reach a trained model HUMAN_REVIEW. Permission is never assumed.
• Using assets at inference time — Runtime/inference use (retrieval, display, generation context) assessed against the reviewed profile independently of training rights.
• Redistributing licensed assets — Passing the asset (or copies) to third parties requires an explicit documented redistribution permission in the controlling terms.
• Modifying and creating derivatives — Modification/derivative-work rights are checked against the reviewed profile; duties such as attribution are surfaced as required next actions.
• Sublicensing rights to others — Granting others rights under the license. Creative Commons and GNU licenses do not allow it; MIT and Apache 2.0 do. a2ally returns BLOCK where the license prohibits it.
• Deploying assets in production systems — Using an asset in a live product or service. Assessed against the license's grant and duties; deploying adapted material counts as sharing it, which NoDerivatives licenses do not allow.
• Attribution duties — Where a profile carries an attribution duty, assessments flag attribution planning as a required next action before use.