. One risk that must be guarded against is that those who develop the superintelligence would not make it generically philanthropic but would instead give it the more limited goal of serving only some small group, such as its own creators or those who commissioned it.
transparency