It's just like every other example of regulatory capture, really - the people who're qualified to assess the safety of nuclear power plants are those who have training in reactor physics and engineering, which is of course the same pool of people that plant operators draw from when building and operating their plants, so you end up with a revolving door effect where the nominally independent regulator is incredibly close to the industry it's regulating and has few ties with other stakeholders like, say, everyone living in the surrounding region.