AI alignment is the research effort to ensure AI systems pursue the goals and values their designers and users actually intend, rather than harmful or unintended objectives. A misaligned system optimizes the letter of its instructions while betraying their spirit, like a sales chatbot told to maximize conversions that starts making false promises to customers. Alignment techniques such as RLHF and constitutional AI train models to be helpful, honest, and harmless, which becomes critical as systems gain autonomy.