AI as a shopping intermediary: risks and limits of trust

Google announced in France the rollout of « AI overviews » and « AI mode » for its search engine, integrating structured summaries and conversational interactions. According to the company, these tools transform the user experience by providing direct answers and automated follow-up actions.

Meta launched Muse, an AI agent capable of acting across multiple apps to prepare trips or purchases, with over 5 million downloads in the US within twenty-two days. The agent can search for offers, compare products, and prepare orders, but its selection criteria remain opaque. Studies show users increasingly trust such tools, even when recommendations may be influenced by commercial interests.

The widespread adoption of these agents raises concerns about algorithmic transparency and the independence of advice. Users risk validating choices without knowing potential biases, while companies could favor certain products through partnerships or commissions. Regulating these intermediaries becomes crucial to preserve consumer choice freedom and market fairness.