Artificial intelligence (AI) systems can be used to promote equality and non-discrimination, but without the right AI governance strategies, they can have the opposite effect. This paper aims to provide policy-makers with the tools they need to build beneficial AI systems. Harmful biases in AI systems often disproportionately affect racialised, low-income and minority groups. For example, algorithms that over-or underestimate scores create false-positive and false-negative results that can perpetuate systemic discrimination. Policymakers can address these concerns through interventions and regulations that promote AI fairness.