Sponsored recommendations pose a trust problem for AI agents

Advertising inside AI agents could conflict with users’ expectation that the agent acts exclusively in their interest. A user objected when free service Instinct sent promotional invitations, viewing the outreach as benefiting the agent rather than the user despite the product’s operating costs.

The analysis considers sponsored agentic actions in ChatGPT and similar products as a possible future advertising format. Companies may pay for recommendations that create incremental demand, but a referral to Amazon is less valuable when the user already intended to shop there. Trust and perceived loyalty could be central constraints on monetizing agents.