Skip to main content
Every result Lightdrift returns carries a rights object. The goal is simple: your agent should never have to guess what it’s allowed to do with an image.

​
The posture

Lightdrift reports rights as declared by the source. We collect the license and permission information a source publishes alongside each image and normalize it into one machine-readable shape. We do not verify declarations or guarantee legal clearance — the same posture as an image search engine, made explicit and structured. That means:
  • Copyright — the license shown is what the source declared. For well-governed sources (CC0, Creative Commons, US-government public domain) declarations are reliable in practice, but they remain declarations.
  • Beyond copyright — a copyright license never covers everything. People in a photo have likeness rights; logos have trademark rights; some buildings and artworks have their own restrictions. Clearing those is on you and your users.
Every rights object carries this posture verbatim in its basis field, so an agent reading a single result has the full context.

​
The fields

​
A practical reading order for agents

  1. Check license against what the task needs (commercial, derivatives).
  2. If attribution_required is true, carry attribution through to the final output.
  3. Treat null permission flags as “the source didn’t say,” not as “no.”
  4. If the subject is a person, logo, or artwork, remember those rights are separate from the copyright license.