A French complaint against Google and the New York Post’s Hamilton launch reveal four competing responses to the economics of the answer layer.
The conflict at the centre of AI search is becoming harder for publishers to ignore. Answer engines need original reporting, specialist analysis and credible evidence. Yet the better those systems become at synthesising information, the less often a reader may need to visit the publisher that produced it.
Two developments on 11 August made that tension unusually clear. In France, the Alliance of General Information Press asked the competition watchdog to act against Google’s AI-generated article summaries. In the United States, New York Post Media Group launched Hamilton, a Gemini-powered news experience designed to keep the answer conversation inside the publisher’s own apps. [1] [2]
These are not opposite stories. They are two answers to the same question: if AI systems depend on publisher evidence, how should publishers recover economic value from the information they produce?
<h2>The economics of the answer layer</h2>
The traditional search bargain was imperfect but intelligible. A publisher created a page, a search engine indexed it, and a portion of the people who saw the result clicked through. The publisher could then monetise the visit through subscriptions, advertising, commerce or a combination of those models.
Generative search changes the path. A publisher still produces the reporting or analysis. An AI system retrieves and synthesises it. But the user may receive a satisfactory answer without leaving the interface. Attribution can still have strategic value, but a citation and a visit are not economically equivalent.
<h2>The French complaint is a traffic signal, not a universal estimate</h2>
Reuters reported that the French Alliance objected to Google’s rollout of AI-generated article summaries without prior authorisation or dedicated remuneration. The Alliance cited an Arcom estimate that publisher-site traffic had fallen by between 33% and 38% because of AI-generated summaries. [1]
That range should not be treated as a universal effect size. It is an estimate cited in a specific French dispute, not proof that every publisher, geography or query category experiences the same decline. It is nevertheless a sharp illustration of the distribution question now confronting news organisations.
If a large share of the economic return from original information disappears, fewer organisations may be able to fund reporting, expert editing, product testing and specialist analysis. That creates a source problem for the answer engines that depend on those materials.
<h2>The Post is building its own answer interface</h2>
New York Post Media Group chose a different response. Hamilton is a free personalised experience inside the New York Post and California Post apps. It combines citation-backed conversational search, personalised digests, tailored recommendations and notifications. The Post says it is powered by Google’s Gemini technology. [2]
Axios reported that the underlying arrangement is an enterprise cloud relationship rather than an AI content-licensing deal. Google Cloud provides infrastructure, models, search and recommendation capabilities, while the Post controls the editorial boundaries and product experience. [3]
The distinction matters. The Post is not merely licensing information to another answer engine. It is using AI infrastructure to create its own interface between its archive, its live newsroom and its audience. The architecture becomes publisher content, publisher-owned answer experience, reader.
<h2>Publishers are choosing four different responses</h2>
Some publishers are confronting the answer layer through regulation. The French complaint argues that AI platforms should not derive commercial value from press content while weakening the traffic economics that fund its creation. Others are pursuing licensing, seeking payment and clearer terms when their material is used to train, ground or improve AI products.
The Publishers Association says AI licensing is an established and growing market for book and journal publishers, with retrieval-augmented generation licensing becoming a meaningful part of the market. That does not settle the economics for news publishers, but it shows that a model beyond referral traffic is forming. [5]
A third response is optimisation. Publishers can structure their journalism for retrieval, citation and subject authority, then measure where their reporting appears inside answer systems. Google’s own June announcement framed this as a website-owner choice, reporting more than 2.5 billion monthly active users for AI Overviews and more than one billion for AI Mode while testing new controls and Search Console insights. [4]
The fourth response is to become the answer interface. Hamilton is one example, but the logic is broader. Once a publisher owns the conversation, it can guide a reader to deeper reporting, subscriptions, related coverage and other forms of value rather than competing only for a click from another company’s answer box.
<h2>The evidence layer is becoming a market of its own</h2>
The future is unlikely to be a simple choice between open crawling and complete exclusion. Microsoft’s Publisher Content Marketplace is designed around publisher-defined licensing and usage terms for specific grounding scenarios, with usage-based reporting. Publishers Weekly reported that Microsoft launched the hub while Amazon was discussing a similar marketplace, although Amazon had not confirmed a specific product announcement. [6]
Standards are also emerging. In February, the Guardian, BBC, Financial Times, Sky News and Telegraph Media Group formed Standards for Publisher Usage Rights, or SPUR, to seek licensing frameworks, technical tools and shared standards for AI use of journalism. [7]
This points to a more complicated evidence economy. Some material may remain openly crawlable. Some may be licensed for grounding. Some may sit behind publisher-controlled interfaces. Each model changes the incentives for creating the original material that answer engines need.
<h2>Why this matters to off-engine AEO</h2>
AEO is often discussed as a property of the brand website: make content accessible, establish clear entities, publish evidence and create useful pages. But answer systems also draw confidence from the broader information environment, including specialist publications, independent reporting, analysts, reviews and other third-party sources.
That makes publisher health an off-engine AEO issue. Brands need credible independent evidence, but those sources need sustainable ways to fund the work that makes their evidence valuable. A citation strategy that ignores the economics of the cited source is incomplete.
<h2>AEO Updates Takeaway</h2>
The publisher fight is not simply a copyright argument. It is a dispute over who controls the interface between information and the audience. Publishers are starting to regulate, license, optimise and build their own answer experiences. For brands, the implication is clear: the third-party evidence layer that shapes AI recommendations is becoming a governed and commercial system, not an unlimited free input. The future of AEO will depend not only on becoming a source AI wants to use, but also on whether the sources AI relies upon can keep producing the information that makes reliable answers possible.
<h3>References</h3>
[1] <a href="https://www.reuters.com/world/french-media-asks-french-anti-trust-watchdog-act-googles-ai-2026-08-11/" target="_blank" rel="noreferrer">Reuters, French press body asks competition watchdog to take action over Google AI, 11 August 2026</a>.
[2] <a href="https://nypost.com/2026/08/11/media/the-post-launches-hamilton-a-new-way-to-find-and-understand-the-news/" target="_blank" rel="noreferrer">New York Post, The Post launches Hamilton, 11 August 2026</a>.
[3] <a href="https://www.axios.com/2026/08/11/new-york-post-hamilton-ai-chatbot" target="_blank" rel="noreferrer">Axios, New York Post launches Hamilton AI chatbot, 11 August 2026</a>.
[4] <a href="https://blog.google/products-and-platforms/products/search/new-controls-website-owners/" target="_blank" rel="noreferrer">Google, New opportunities, control and insights for website owners, 3 June 2026</a>.
[5] <a href="https://www.publishers.org.uk/publications/content-superpower-uk-publishing-ai-licensing-market/" target="_blank" rel="noreferrer">Publishers Association, Content Superpower: UK publishing and the AI licensing market, 3 March 2026</a>.
[6] <a href="https://about.ads.microsoft.com/en/blog/post/february-2026/building-toward-a-sustainable-content-economy-for-the-agentic-web" target="_blank" rel="noreferrer">Microsoft, Building Toward a Sustainable Content Economy for the Agentic Web, 3 February 2026</a>; <a href="https://www.publishersweekly.com/pw/by-topic/digital/content-and-e-books/article/99670-microsoft-amazon-plan-ai-content-marketplaces.html" target="_blank" rel="noreferrer">Publishers Weekly, Microsoft, Amazon Plan AI Content Marketplaces, 10 February 2026</a>.
[7] <a href="https://www.theguardian.com/media/2026/feb/26/guardian-joins-media-coalition-to-protect-original-journalism-from-unpaid-use-by-ai" target="_blank" rel="noreferrer">The Guardian, Guardian joins media coalition to protect original journalism from unpaid use by AI, 26 February 2026</a>.