Recent research has shown that annotations are useful for representing access restrictions to the axioms of an ontology and their implicit consequences. Previous work focused on computing a consequence's access restriction efficiently from the restrictions of its implying axioms. However, a security administrator might not be satisfied since the intended restriction differs from the one obtained through these methods. In this case, one is interested in finding a minimal set of axioms which need changed restrictions. In this paper we look at this problem and present algorithms based on ontology repair for solving it. Our first experimental results on large scale ontologies show that our methods perform well in practice.