A new study offers a different view on the claim that “AI agents are already spending money online on their own.” According to the report, real-world AI agents currently do not complete online payments at scale and autonomously in the way some promotional descriptions suggest.
Market interest in this topic mainly stems from the rapid development of technologies such as AI agents, automated execution, and smart wallets. Some demos do show agents autonomously initiating operations or even payments, which can easily create the impression that AI can already spend money on behalf of users. However, the study argues that such behaviors are not yet widely adopted in practice, and many scenarios remain in experimental or controlled environments.
The research also cautions that for AI agents to truly take on payment functions, issues such as permission control, security, compliance, and attribution of responsibility still need to be resolved. Therefore, the narrative of “AI spending money” should be treated rationally: acknowledge technological progress while avoiding mistaking staged demonstrations for widespread reality.
The original article is a Decrypt report titled “AI Agents Spending Money Online? New Research Says Not Really,” and its content has been paraphrased.
Source and Editorial Note
This article extracts its topic from the public original article and has been rewritten and structurally edited without adding unverified facts. The original author’s views do not represent the position of this site; for dynamic data, please refer to the original page.
