AI answers are cutting publisher traffic, hitting small sites hardest, while courts and licensing deals are still deciding what is owed for the work inside those answers.
- Small sites lose most. Chartbeat reports search referrals down 60% for small sites and 22% for large publishers.Chartbeat data, reported by Axios, 17 Mar 2026, read at source 10 Sep 2026: “small sites lost 60 percent of their search referral traffic, medium sites lost 47 percent, and large sites lost 22 percent”.
- Chatbots send little traffic. AI chatbots bring under 1% of publisher referrals, though that share is growing fast.Chartbeat, via Axios, 2026, read at source 10 Sep 2026: “chatbots still account for less than 1% of all publisher page view referrals”.
- Weaker pages mean weaker answers. If paying for original pages stops working, the material AI answers are built from degrades.
- How data was obtained matters. A US ruling allowed training on lawfully acquired books, not keeping pirated ones.
- Readers can help. Visiting sources on purpose is free, and it is the part you personally control.
The part that decides who survives
The headline number hides the finding that actually matters: the loss is not distributed evenly, and it is inverted. The smaller the publisher, the harder the hit — and the pages being written for machines rather than people are a separate symptom of the same shift.
Search referral decline by publisher size. Chartbeat, Navigating the New Traffic Landscape (Q1 2026 report, network data to 9 Feb 2026), read at source 23 Sep 2026: “Small publishers face a 60% decline in search traffic compared to just 22% for large sites”. Single-source figure; treat the pattern as firmer than the precise values.
Large organisations absorb a 22% loss, sign a licensing agreement, and continue. An independent site losing 60% of its referrals does not have a legal department, a licensing negotiation, or a subscription base to fall back on. The mechanism that concentrates the web is not censorship. It is arithmetic.Chartbeat, Navigating the New Traffic Landscape (Q1 2026 report, network data to 9 Feb 2026), read at source 23 Sep 2026: “Small publishers face a 60% decline in search traffic compared to just 22% for large sites”.
The size of a site shapes how hard AI answers hit it: the smaller the publisher, the bigger the measured loss and the fewer ways it has to absorb it.
What the numbers do not say
This page would be dishonest without these, and most coverage of this topic omits them.
Zero-click rates have not risen monotonically. One tracker measured the zero-click rate on AI-Overview keywords falling through 2025 — on the same keywords before and after an AI Overview appeared, from 33.75% to 31.53% — the opposite of the expected direction. Users may be adapting rather than surrendering.Semrush, AI Overviews Study, refreshed 15 Dec 2025, read at source 23 Sep 2026: “When we analyzed the same keywords before and after they started showing AI Overviews, we found the zero-click rate decreased from 33.75% to 31.53%.”
AI referrals are growing fast. ChatGPT's outgoing referrals rose roughly 52% year over year (September to November 2025). The absolute number is still tiny, but a small number compounding quickly is different from a dead channel.Similarweb data, reported by Digiday, 22 Dec 2025, read at source 23 Sep 2026: “From September to November 2025, ChatGPT referrals increased 52% year over year.”
Licensing money is real. Publishers now report meaningful revenue from AI licensing deals — a line item that did not exist three years ago. It is unevenly distributed, but it is not nothing.
Read the traffic figures as a trend with caveats, not a final verdict: one measure moved the other way, and AI referrals and licensing money are small but real.
Why this is not just a publisher problem
It is tempting to file this as an industry squabble. It is not, for a reason that closes a loop with the rest of this site.
AI answers are made of pages. Every summary an answer engine produces is assembled from material somebody researched, wrote, fact-checked and paid for. If the economics of producing that material collapse, the input degrades — and the answers degrade with it, for everyone, including the people who never click a link.
Add the other measured trend: roughly half of new articles are now primarily machine-generated, per the figures in the quarterly record. Human-made pages are getting less economically viable at the same moment machine-made pages are getting cheaper. That is not a conspiracy; it is two incentives pointing the same way.
If answering a question no longer requires visiting the page that knows the answer, who pays the person who found it out? Licensing deals answer this for a few dozen large organisations. Nothing currently answers it for the independent site, the specialist blog, the forum thread or the documentation page — which together are most of what makes the web worth searching.
This reaches people who never click a link: AI answers are built from pages someone paid to research and write, so if producing those pages stops paying, the answers get worse too.
The courts, and the distinction that actually emerged
The other half of this economy is being decided in litigation, and the early rulings did not land where either side expected.
The question turned out not to be "is training copying?" It turned out to be how the material was obtained.
In the most detailed US ruling so far, a judge held that training a model on lawfully acquired books was transformative fair use — while retaining a corpus the company knew to be pirated was not, and exposed it to statutory damages. That case settled for approximately $1.5 billion, the largest copyright settlement on record in the United States, working out to roughly $3,100 per book. By the March 2026 claim deadline, 440,490 of 482,460 eligible works had been claimed — a 91.3% rate.Bartz v. Anthropic (N.D. Cal.), merits ruling June 2025, settlement announced August 2025, claim data March 2026, read at source 10 Sep 2026 — Judge Alsup held that using books without permission to train AI “was fair use if they were acquired legally” but denied summary judgment on piracy. The settlement covers “almost 500,000 pirated works” at “approximately $3,000 per class work”, and Anthropic must destroy its copies. Final approval was still pending after a May 2026 fairness hearing at the time of writing.
Around it, the picture is genuinely mixed rather than settled. Another judge granted summary judgment to a different lab on fair use, finding the training highly transformative but the plaintiffs unable to show market harm. A third court found against fair use where an AI product competed directly with the database it had been trained on — the first final judgment in this area, now under the first appellate review, argued in June 2026. A German court rejected the equivalent defence outright under its own law. The US Copyright Office has stated that fair use cannot simply be presumed for AI training.
Provenance became the asset. Labs that can document a clean licensing chain for their training data are surviving motions and settling on better terms; those with poorly documented or piracy-tainted pipelines are not. That is why the licensing market appeared so quickly — nine-figure deals between major publishers and the largest labs, plus a new layer of intermediaries aggregating smaller publishers to license collectively.
And there is a feedback loop worth noticing: every disclosed licensing deal makes the fair-use defence harder for the next defendant, because it demonstrates that a licensing market exists and could have been used. The industry's own deal-making is narrowing its own legal argument.
Which connects directly to everything above. The traffic collapse and the licensing settlements are the same question asked twice — once by the market and once by the courts: if a publisher's work ends up inside an answer, what is owed, and to whom? Neither forum has finished answering, and the outcomes will decide whether independent publishing has an economic basis at all.Reasoning, August 2026 — this is the site’s own summary of a fast-moving and jurisdiction-dependent area, and it is not legal advice.
So far the courts have cared less about whether training counts as copying than about how the material was obtained, and the law is still unsettled and differs between countries.
AI answer engines are genuinely useful, and this site is built to be readable by them — its own llms.txt exists for exactly that reason. The problem is not summarisation; it is that the value moved without the payment following. The fix most likely to work is not blocking crawlers, which mostly harms the blocker, but attribution with weight: citations that carry traffic, licensing that reaches beyond the largest players, and readers who deliberately visit sources. That last one is free, and it is the only part you personally control.