No scraping, no brokers, no platform customers. Every record traces to a named business that licensed it to us directly.
Licence chain
Every corpus traces to a named business that signed a training-only licence with Enjamb Labs. The chain is disclosed to a buyer under NDA, to the extent needed to establish provenance.
Training and evaluation only
Buyers may train and evaluate. They may not resell, redistribute or attempt re-identification.
Never platform data
Data from customers of the Enjamb platform is held in separate systems under separate agreements and never enters this pipeline.
Method document
Per dataset: Safe Harbor or Expert Determination, who signed it, the identifier classes covered, the date-shift design, the k used on quasi-identifiers, the tier of each field, and the red-team result.
Exclusions annex
What the seller carved out, and the hashes that prove it never entered the pipeline. Categories, never content.
Watermark and fingerprint
Each delivery is unique. A file found where the licence does not allow it traces to the buyer it was delivered to.
Jurisdiction
Raw records are stored in the United States only, encrypted and access-logged. Every buyer and transfer is screened against US bulk-data rules. Health and financial-account corpora go to labs in the United States and allied jurisdictions only.
Refresh and retraction
Retracted records are removed from the next delivery and listed in the change log.