According to the source, Microsoft tested hundreds of AI agents in a simulated marketplace as buyers and sellers and found they failed basic consumer tasks that humans routinely complete. Source: user-provided social media post referencing Microsoft. Reported failure modes included falling for scams, mismanaging budgets, and not completing intended purchases, signaling reliability gaps for autonomous shopping assistants. Source: user-provided social media post referencing Microsoft. For traders, this suggests near-term execution risk for autonomous AI commerce and potential delays to monetization for agent-driven retail flows, warranting tighter risk controls on AI-commerce exposure. Source: independent analysis based on the provided source. In crypto markets, weaker confidence in autonomous agents can pressure AI-narrative tokens and decentralized commerce themes while shifting interest toward identity, reputation, and escrow primitives that mitigate agent risk. Source: independent analysis based on the provided source.