A new paper on arXiv, Standing, Not Truth: Who May Say What About Whom in Agentic Selection, addresses a growing problem in automated decision-making. Software agents now routinely choose among competing products or services on behalf of users, but the descriptions of those offerings come from the parties who stand to gain from the selection. This creates a question of trust and authority: whose statements should an agent rely on?
The abstract outlines a specific scenario: a lawful participant, publishing under its own identity, compromising no channel and issuing no instruction, can influence the outcome. The paper appears to explore the limits of such influence, asking what counts as legitimate speech in an agent-mediated marketplace.
Because the abstract is cut off, the full argument is not yet visible. What is clear is that the authors are probing the boundary between truthful description and strategic positioning, and how agents should weigh competing claims when the speakers have skin in the game.