Both the American Geophyical Union and PLOS have recently announced new policies dealing with the data that underlies published research. The PLOS statement recognized that having open access literature without providing the underlying data was incongruous, and both policies provide recommendations for what to share, and how to share it.
In both statements, the requirement is not necessarily for all research data to be shared, but for the critical parts of data that were used to complete the published research and analysis. The two groups also require that data is made easily accessible, within legal constraints, and not held by the authors as the sole gatekeeper to the data. The expectation is for the data to be stored in trusted repositories: AGU provides a link to suggested repositories; PLOS suggests authors use repositories verified with criteria such as that provided by the Centre for Research Libraries or Data Seal of Approval (perhaps recognizing that when public archives aren't used, access to the data rapidly diminishes over time).
Recognizing that there are a variety of reasons all data may not be able to be shared publicly, each policy provides options for privacy or legal concerns, but requires that a statement be made to that effect, explaining why part or all of the data can not be shared.
With funding agencies and the federal government moving swiftly toward open access for all research outputs, and the proliferation of tools making it easy to share data, more groups and publishers will no doubt be adopting similar policies in the very near future.