Create a documented public view while preserving controlled originals and accountable reanalysis.
Sensitivity depends on context
Sensitivity can depend on breeding sites, season, collection pressure, land access and community agreements, not only a species name. GBIF guidance addresses information whose release could cause harm. Maintain a decision register covering taxa, region, release scope and review date. Reassess classifications as threats and conservation circumstances change.
Generalize deliberately and document the change
Separate precise research records from public aggregates, using an appropriate spatial unit when needed. Document the released location and its uncertainty rather than silently removing decimal places. Place names, filenames, image metadata, map tiles and downloads can reveal locations even when a visible map is blurred. Review the entire release package.
Make access accountable
Assign access according to role and research purpose, with approvals, expiry, usage conditions and download history. Short-lived download links and audit records are possible implementation choices. External AI services should remain within the same authorization boundary. Convenience should not silently broaden the data-sharing terms of the underlying research.
Preserve reproducibility without exposing every coordinate
Methods, model versions, aggregates, generalization procedures and validation summaries can often be shared while precise originals remain controlled. Provide an appropriate route for authorized reanalysis and explain what analytical capability is lost through generalization. State the limits of independent verification instead of implying that either unrestricted release or complete secrecy is the only option.
Evidence & further reading
Check the original study conditions, species and location before using a result in a plan. Do not transfer findings universally across species and settings.