Financial Market Fragility in the Era of AI Planning

Joint with Itay Goldstein, Yan Ji

November 2025

This paper studies how AI planning, the core technology behind Agentic AI systems that reason over sequences of actions toward long-term goals, affects financial market stability. We develop a dynamic trading framework with positive-feedback investors, constrained arbitrageurs, and oligopolistic informed speculators who may coordinate through intertemporal planning: trading aggressively to generate (negative) bubbles and unwinding gradually to extract profits. Such coordination differs from traditional collusion because it faces two unique challenges: time inconsistency, as coordinated plans are ex post incentive-incompatible, and weak punishment, as deviations are difficult to penalize when no large (negative) bubble ultimately materializes. We characterize the equilibria featuring planning-based coordination among speculators. In simulation experiments, AI-planning speculators trained via reinforcement-learning algorithms with explicit planning modules autonomously discover and implement intertemporal collusive trading strategies based on compounded price-trigger rules, coordinating without communication or intent. With strong feedback trading, AI-planning speculators coordinate dynamically on destabilizing strategies that create and exploit (negative) bubbles, manipulating feedback traders and amplifying market fragility.