Statement from engineering -

In C# is required to handle conversion of parameters when it calls across to a native API. Certain basic parameters need no special handling, but more complicated data, such as structures and arrays require further work. If you *cannot* change the interface to a native function and are stuck with certain data types which cannot be safely marshalled - the "unsafe" keyword comes in. It enables you, as a developer, to accept the "unsafe" data interface - but note that in turn this marks your whole application as "unsafe" too. This is generally not what a publisher wants.