Algorithms are increasingly prevalent in industries such as e-commerce, enabling rapid price adjustments. We show that reinforcement learning algorithms, widely used in dynamic pricing and retail optimization, can independently engage in predatory pricing without any human guidance. When aggressive price competition is too costly, these algorithms may instead exploit the learning dynamics and rigidity of rivals’ pricing systems, shaping competitors’ early learning outcomes in ways that sustain high prices after entry. As a result, competition may be softened and entry less effective, despite the absence of explicit exclusionary or manipulative intent. These outcomes emerge from standard algorithmic design choices rather than deliberate strategy and may elude existing regulatory frameworks. We discuss managerial safeguards and regulatory responses aimed at preserving competition in increasingly AI-driven retail markets.
Abada et al. (Sun,) studied this question.