Possibility, with responsibility.

Start with authority and provenance

We assess who created or holds a source, who can authorise disclosure and what uses are permitted. Ownership, confidentiality, AI use, commercial reuse, transformation and onward licensing each need evidence. A checkbox is a statement for review, never automatic approval.

Separate the decisions

A commercially interesting offer is not an approved dataset. Commercial status, rights review and privacy assessment remain separate. Human reviewers record the decision and date; automated classification must not fill missing permissions.

Use proportionate review

Health, finance, education, children, precise location, biometrics, private communications and security-sensitive information require enhanced review. We request descriptions initially. Any samples need a separately agreed, access-controlled process.

Know what is outside the route

Stolen or leaked data, credentials, unauthorised personal dossiers, covert tracking products and content without a lawful supply route are not acceptable. We do not market personal contact lists or private personal records.

Privacy is more than removing names

Removing direct identifiers does not establish effective anonymisation. Residual identification risks, source-country requirements and intended use need appropriate assessment. Specialist advice may be required before higher-risk processing or supply. Synthetic data also needs generation rights, clear labelling and documented limitations.

Raise a concern

Use the privacy route to request review, correction or removal. Share only enough information to describe your concern. We will not publicly confirm whether an individual appears in a dataset or promise removal from an already trained model.

Supplier content is not sent to external AI tools without explicit authority and appropriate terms.