Data ReserveAI data licensing research
Section 12 / Next steps

How should we establish an eligible subset?

Data Reserve research deskSection 12Page 3 of 13Last updated 4 November 2026

Ask the archive owner and counsel to group material by source and rights history: staff work, freelancers, syndication, acquired collections, licensed images, third-party tables and other relevant categories. Link a representative document to each cohort, while recording that one sampled contract does not establish every item's status.

Contributions may have separate owners under section 201. Resolve the specific grant rather than relying on an archive label. Circular 30 is the official background for the work-made-for-hire distinction.

Produce a register with: cohort, records included, source of authority, proposed use, restrictions, unresolved question, reviewer and release status. Use statuses such as “not yet reviewed,” “excluded,” and “approved for this named evaluation subject to conditions.” Avoid a universal “AI cleared” label.

If the collection contains personal or confidential material, assign a separate review. The FTC's AI privacy and confidentiality guidance stresses that businesses must honour their commitments; a copyright licence does not answer every disclosure question.

Advance when: the intended next disclosure has a documented scope, an eligible subset and accountable approval. A full-archive legal review need not be completed merely to discuss non-confidential metadata.

Sources cited on this page

  1. section 201 ↗copyright.gov
  2. Circular 30 ↗copyright.gov
  3. AI privacy and confidentiality guidance ↗ftc.gov