Evidence since mid‑2024 shows the emergence, and growing scale, of specialized intermediaries that clear or license publisher/creator content for AI systems, including answer/search products:

• The long‑standing Copyright Clearance Center (CCC) created new AI‑specific collective licenses in 2024–2025 that let organizations ingest and use large corpora of copyrighted works in AI systems via a central license, and then expanded this to an AI Systems Training License for models whose outputs are used externally—i.e., commercial, customer‑facing generative AI products. CCC acts as an intermediary between thousands of publishers and AI developers, and reports strong publisher uptake. (grokipedia.com)
• In the U.K., the Copyright Licensing Agency (CLA), Authors’ Licensing and Collecting Society (ALCS), and Publishers’ Licensing Services (PLS) are rolling out a collective licensing framework for AI training and retrieval‑augmented generation (RAG), explicitly designed so AI developers can get permission to use a wide range of text‑based published works through a single license, with CLA administering fees and distributions to rightsholders. This is exactly an intermediary model for AI training/RAG at national scale. (ailawandpolicy.com)
• A Bloomberg Law report describes a wave of “AI licensing startups” that aggregate and license copyrighted works for AI, including Created by Humans (authors and book publishers licensing to AI systems), Narrativ (voice actors’ likenesses), and ProRata, whose tools decompose AI outputs and route compensation back to contributors. These are explicitly characterized as "middlemen startups" aggregating many rightsholders into a single licensing package. (news.bloomberglaw.com)
• Separate coverage of ProRata AI shows it has enrolled 400+ (later 500+) publications—among them major brands like the Financial Times, Axel Springer titles, The Atlantic, Fortune, Vox Media, The Boston Globe, and others—into a 50/50 revenue‑share model. ProRata analyzes generative‑AI outputs, attributes them back to source publishers, and distributes revenue accordingly, functioning as a rights‑clearing and payment intermediary between publishers and AI services. (axios.com)
• ProRata also operates Gist.ai, an AI‑powered search/answer engine that uses this licensed publisher content to generate responses, a concrete example of an AI‑answer product built on top of a specialized rights‑clearing intermediary rather than purely on unlicensed web scraping. (digiday.com)

While Google specifically has not publicly outsourced AI Overviews licensing to a single neutral clearinghouse, the broader substance of the prediction—that a new class of intermediaries, whose core business is clearing/licensing publisher and creator content for AI answer/search and other generative uses, would emerge and reach meaningful scale—has clearly materialized by late 2025. The combination of large, AI‑specific collective licenses (CCC, CLA/ALCS/PLS) and scaled startups like ProRata/Created by Humans matches the predicted pattern closely enough to count this as right.