Embedded undisclosed product placement in an experimental LLM chatbot was difficult to recognize and shifted product evaluations while reducing benchmark answer quality by no more than three percent.

What this does not establish

Laboratory product judgments are not clicks, purchases, long-run welfare, or evidence that hidden advertising is legally or ethically acceptable.

Counterevidence & uncertainty

Participants could prefer hidden-ad responses even while embedded persuasion creates a disclosure and autonomy risk.

What would change the reading

Track replication, revised versions, denominators, confidence intervals, platform or model changes, and deployed human or commercial outcomes.

Primary routes

1. AIAO-S004https://arxiv.org/abs/2409.15436Open source record →