We introduce learning based on genetic algorithms in a principal-agent model of optimal contracting under moral hazard. Applications of this setting abound in finance (credit under moral hazard), public finance (optimal taxation, information-constrained insurance), development (sharecropping),...