The law of state responsibility codified by the International Law Commission (ILC) in its Articles on the Responsibility of States (ARS) is analysed with respect to unlawful acts of an armed group where: (i) the state provides substantial support to the group; (ii) the group exercises governmental authority over part of the state’s territory; and (iii) a successful insurrection results in the establishment of a new government or a new state.

Part One addresses (i) and the control and agency paradigm central to the ARS. The ‘effective control’ test applied to identify a state’s agents (article 8 ARS) means that state-sponsors of terrorism or international crimes will avoid direct responsibility by holding proxies at arms-length. This thesis challenges the arguments that attribution should be based on ‘overall control’ in such cases. Instead, this thesis proposes a rule of derivative responsibility for complicity that draws on primary obligations under international humanitarian and human rights law that prohibit encouragement or facilitation of certain acts.

Part Two addresses (ii) and (iii) and the application of articles 9 and 10(1) and (2) ARS respectively – exceptions to the control and agency paradigm that have been generally overlooked in discourse. With regard to (ii) this thesis counters recent scholarship that argues that article 9 will apply to the ‘governmental’ acts of armed opposition groups. Doctrinal analysis shows that article 9 will only cover the conduct of an armed group that acts in the interests of, or with the acquiescence of, the state’s authorities. A critical approach is taken to the ILC’s codification of articles 10(1) and (2) that cover circumstances (iii): Contrary to the ILC’s position, this thesis argues that the ‘peace v. justice’ debate should not influence article 10(1)’s application to governments of national reconciliation. Further, it is shown that article 10(2) is seriously flawed: the rule requires the new state to be bound by international obligations before that state existed in fact, a notion not supported by state practice or doctrine.