Before releasing a dataset, it is important to ensure that access to the dataset is appropriately configured. By default, new datasets are public at the overview access level, and must be explicitly shared with others in order for them to get metadata or data access.
In order to change a dataset's access configuration, click the Configure access button on the dataset editor.
For every access level, you may use one of three paradigms to control access. Different levels can be controlled by different paradigms — for example, you may want anyone to be able to see overview access, specific people to have metadata access, and only the subset of those specific people who have completed certain requirements to have data access.
If a dataset is public at a particular access level, anyone on the internet can access the dataset at that level (once it's released).
For organization datasets, administrators may specify that only organization members may access the dataset at that level.
Direct access allows the data owner to specify the individual users who can access the dataset at that level.
Available only to datasets owned by an organization. Users must complete (and be approved for) all requirements for an access level in order to gain access.
Learn more in the requirements reference.
Organizations have the option to codify their access configurations in permission groups, which can then be applied across datasets.
Learn more in the permission groups reference.
When you first create a dataset within an organization, it is unpublished. Unpublished datasets can only be viewed by organization administrators and assigned dataset editors. Once a dataset is published, it can be "unpublished" at any time to restrict access to administrators only, locking down the dataset without having to modify its configuration.
Organizations can also configure various usage restrictions on a dataset — for example, to limit exports to specific compute environments.
Learn more in the usage restrictions reference.