A 3,000-Publisher Collective Just Signed Its First AI Customer. Yahoo Scout Will Pay for the Read.
On 12 November, more than 3,000 media brands, content owners and publishing platforms launched the Internet Licensing Collective (ILC), a nonprofit that licenses publisher content to AI companies under a single agreement (named-party figure, ILC). Its first commercial licensee is Yahoo Scout, Yahoo's AI answers product.
The headline is not Yahoo. The headline is the mechanism. An AI answer engine has agreed to pay a pool of publishers it never negotiated with one by one, for uses that include grounding answers in real time. That is a payment for the read, not for the referral.
What is the Internet Licensing Collective?
The Internet Licensing Collective is a nonprofit collective rights organisation, steered by a publisher committee, that sells one blanket licence to AI companies and distributes the revenue to participating publishers. It was previously called the RSL Collective, and it co-authored Really Simple Licensing (RSL), the open, machine-readable standard publishers use to state licensing terms for AI.
Collective licensing is not new. Music has run on it for decades: a broadcaster or streaming service pays a collecting society once, and the society splits the money among rights holders by usage. ILC applies the same model to the web, and its launch release explicitly calls this a "Spotify moment for the internet".
How does the money actually move?
According to ILC's launch release, the model has four parts.
One licence, chosen by use. AI companies license only the uses they need. The release lists training, retrieval-augmented generation (RAG), grounding, agentic applications and AI-generated overviews as separate categories.
Publisher control. Publishers join free, choose which licences and which licensees they accept, and keep any direct deals they already have. Participation is non-exclusive.
Allocation by value, not catalogue size. Revenue is split under published Allocation Rules that weigh how content is used, the economic value of the AI product, and how often a publisher's content is relied on. ILC says this is designed so that a small specialist publisher that materially improves an answer is paid fairly against a large generalist.
Nonprofit costs. ILC projects operating costs at or below 10% of revenue (named-party figure, ILC). The rest passes to rights holders.
What the release does not say is what Yahoo Scout is paying, on what basis, or how much content it is licensing. No fee, rate or minimum has been published.
Why does Yahoo Scout matter, and why does it not settle anything yet?
It matters because it is a working contract, not a standard waiting for adoption. When RSL launched in 2025 with more than 1,500 supporters (named-party figure, ILC), Yahoo was among them, but no AI company had agreed to honour it. A signed licensee turns a protocol into a payment route.
It does not settle anything because Yahoo Scout is a new product, and AI answers are dominated by ChatGPT, Google's AI Overviews and AI Mode, Gemini, Perplexity and Copilot. None of those has joined. Until one of them does, the collective prices the read for one buyer and leaves the largest readers outside the licence.
How big could collective AI licensing be?
Alongside the launch, Mather Economics published The Price of Intelligence, which ILC describes as an independent analysis that it did not commission or fund. Its central estimate is that AI companies have an economic willingness to pay between 15% and 35% of their revenue to keep access to licensed content (named-party figure, Mather Economics).
Applied only to the current revenue run-rate of the two largest frontier model companies, Mather sizes the collective licensing market in the tens of billions of dollars, and says it could overtake total 2025 US display advertising revenue within three years (named-party figure, Mather Economics).
Treat this carefully. It is a willingness-to-pay model, not money paid. The benchmark it leans on is music: the release notes that Spotify pays close to 70% of revenue to rights holders and YouTube shares 55% with creators (named-party figures, ILC release). The measured flow from AI answers to most publishers today is close to nothing.
What does this mean for publishers?
Three things.
First, collective licensing is the first route to AI revenue that does not require a publisher to be big enough to negotiate. That is the long tail's real problem, and ILC is built for it.
Second, the allocation rules are the product. Payment by "value contributed" means someone has to measure how often each publisher's content is relied on in an answer. Publishers should ask how that is measured and audited before they assume a fair share.
Third, joining is free and non-exclusive, so the downside is low. The question is not whether to join. It is what covers the reads a licence does not.
What does this mean for advertisers?
The AI answer is becoming a licensed, metered product. Once content inside an answer is counted per use, the answer starts to carry the same plumbing as any other media: a measured unit, a price and a payer.
For brands, that has two implications. The sources an engine licenses will shape what it says about your category, so knowing which publishers sit inside an engine's licence is part of knowing how visible you are in its answers. And metering is the precondition for paid, measurable placement in and around answers. Display, paid search and social do not appear inside an AI answer. Whatever does will be built on infrastructure like this.
The longer arc
Publishers' response to AI has moved in three steps. First they blocked. Then the largest signed bilateral deals with the big AI companies. Now the rest are pooling rights to sell at internet scale. Each step moves from treating AI as a threat to keep out towards treating the read as inventory to price.
The music parallel is useful but incomplete. Streaming services needed a licence to operate at all. AI answer engines can still read most of the open web without one.
the blankspace read: A licence prices the read at the platform's front door, and only for platforms that walk through it. Today that is one answer engine. Every Live Search Agent dispatched by an engine that has not signed still reads the page, runs no JavaScript, fires no ad tag and pays nothing. Collective licensing and per-read monetisation are not rivals: one is a contract, the other is a meter on the reads that arrive without one. Publishers will need both. Monetise the read, not the referral.
Key figures
| Figure | Value | Provenance |
|---|---|---|
| ILC participants at launch | More than 3,000 media brands, content owners and publishing platforms | Named-party figure (ILC) |
| First commercial licensee | Yahoo Scout | Named-party figure (ILC, Yahoo) |
| Launch date | 12 November 2026 | Named-party figure (ILC) |
| Projected ILC operating costs | At or below 10% of revenue | Named-party figure (ILC) |
| RSL supporters at 2025 launch | More than 1,500 | Named-party figure (ILC) |
| AI companies' willingness to pay for licensed content | 15% to 35% of revenue | Named-party figure (Mather Economics, via ILC release) |
| Estimated collective licensing market | Tens of billions of dollars | Named-party figure (Mather Economics, via ILC release) |
| Spotify payout to rights holders | Close to 70% of revenue | Named-party figure (ILC release) |
| YouTube share with creators | 55% | Named-party figure (ILC release) |
| Yahoo Scout fee or rate | Not disclosed | Reported (absent from the release) |
Primary source, opened and verified: The Internet Licensing Collective (ILC) Launches with Yahoo Scout as First Commercial Partner, ILC press release, 12 November 2026