Agentic AI and hallucinations

We model a competitive market where AI agents buy answers from upstream generative models and resell them to users who differ in how much they value accuracy and in how much they fear hallucinations. Agents can privately exert effort for costly verification to lower hallucination risks. Since interactions halt in the event of a hallucination, the threat of losing future rents disciplines effort. A unique reputational equilibrium exists under nontrivial discounting. The equilibrium effort, and thus the price, increases with the share of users who have high accuracy concerns, implying that hallucination-sensitive sectors, such as law and medicine, endogenously lead to more serious verification efforts in agentic AI markets. • We study competition among AI agents choosing model and verification effort. • Our model links user mix, patience, and AI service reliability. • Relational contracts incentivize AI agents to verify answers and curb hallucinations. • Market composition disciplines agents and resolves the moral-hazard problem. • Verification effort rises with the market share of high-stakes users.