Algorithmic bias occurs when machine learning models perpetuate or amplify human prejudices, data imbalances, or proxy discrimination through their predictions and classifications. This creates legal liability under anti-discrimination laws, reputational damage, and ethical concerns. Amazon discontinued an AI recruiting tool in 2018 after discovering it penalized resumes containing the word 'women's,' having learned bias from historical male-dominated hiring data. Managing this risk requires diverse development teams, bias testing throughout the model lifecycle, fairness metrics, human oversight of high-stakes decisions, and transparency about how automated systems make determinations.