[Paper Review] AI Search: Media Beyond a Tool

The author begins by defining AI search not as a technical tool for finding information but as "media" exercising social influence. Citing Marshall McLuhan''s insight "the medium is the message," the author argues that the manner in which AI search delivers information ("how") is itself restructuring human senses and social order. Traditional internet search was a map directing people toward information; AI search has become the agent that directly composes information content and forms meaning before delivering it to users — making it impossible to view AI search operators as neutral technical providers, creating the logical foundation for requiring media-equivalent public responsibility. RAG technology''s paradigm shift: traditional search engines served as mere conduits (pipes) to information, qualifying for Online Service Provider (OSP) liability exemptions. But Retrieval-Augmented Generation (RAG) involves AI directly summarizing and reconstructing retrieved external information — a selection and editorial process that transforms the operator from intermediary to active editor. The copyright dimension: AI search using copyrighted news content for answer generation without directing traffic to source sites raises both copyright infringement and market substitution questions. The author proposes distinguishing between "informational use" (permissible learning) and "expressive use" (generating content that substitutes for the original source''s market). Regulatory implications: AI search operators exercising editorial judgment (selecting which sources to summarize, how to present conflicting information, what context to include) should bear responsibilities analogous to media organizations — including accuracy obligations, correction duties, and transparency requirements — rather than benefiting from the neutral platform liability shields designed for passive conduits.