This paper revisits optimal insurance design under both risk and ambiguity using an optimal control framework, jointly solving for the indemnity function and the premium. Extending the smooth ambiguity model of Klibanoff et al. (2005), the authors prove the existence of an optimal policy for a risk-averse policyholder and a risk-neutral insurer, both of whom may be ambiguity-averse or ambiguity-neutral. The paper provides a characterization of how risk and ambiguity are shared between the insurer and the policyholder. It shows that under one-sided ambiguity aversion, a straight deductible is not optimal due to the resulting incompleteness of the contract.