Short answer. The citation graph is being renegotiated commercially and legally at the same time, and brands are not party to either process. OpenAI has assembled roughly 20 publisher partnerships covering 160+ outlets. Perplexity pays 80% of one subscription tier's revenue into an initial $42.5 million publisher pool. As of 31 May 2026, nine organisations had active suits against Perplexity — several of them among the most-cited sources in AI answers. Your citation mix can therefore move because of a contract or a ruling, with nothing on your site having changed.
Most AI visibility planning treats the set of sources an engine draws on as a fixed feature of the system, to be competed against on content quality. It is not fixed. It is a commercial arrangement under active litigation, and the terms are being written now by parties who have no reason to consider a brand's interests.
This piece sets out the three mechanisms at work, then what actually follows for a company that is not a publisher.
Three things happening at once
| Mechanism | What it does | Who benefits | What it means for a brand |
|---|---|---|---|
| Licensing deals | Paid access and preferential treatment for named publishers | Large news organisations | Structurally advantaged competitors for citation slots |
| Revenue share | Pays publishers per citation event | Enrolled publishers, including smaller ones | A market price now exists for a citation |
| Litigation | Contests unlicensed use through the courts | Rights holders with resources | Uncertainty about which sources remain available |
None of the three is open to an ordinary company. All three change the field it operates on.
What does the licensing layer look like?
OpenAI has assembled roughly 20 publisher partnerships covering more than 160 outlets in over 20 languages, per the LLM Pulse licensing tracker. Perplexity's revenue-share programme has added the Los Angeles Times, Adweek and The Independent alongside Time and Fortune.
The Perplexity structure is the more interesting of the two, because it makes a citation a unit of account. Its Comet Plus subscription pays 80% of its revenue to participating publishers, against 20% retained for compute, from an initial pool of $42.5 million. Launch partners include Condé Nast titles, Fortune, The Washington Post, Los Angeles Times, Le Monde and Le Figaro. Announced in late August 2025 with user access from early October 2025, payouts derive from three things: direct traffic to publisher sites, citations within answers, and assistant usage during task completion.
Once a citation has a price, three things become sayable inside a company that were previously hand-waving: a citation has a market value, that value is set by the engine rather than by the brand, and being cited is worth something even when nobody clicks. That is the strongest available external evidence for the AEO business case — with the caveat that it prices publisher content, not brand content.
Who is suing, and why does it matter to you?
As of 31 May 2026, nine organisations had active suits against Perplexity over alleged copyright or trademark infringement, including CNN, The New York Times, News Corp, Encyclopedia Britannica and Reddit.
Note who is on that list. Reddit and Encyclopedia Britannica are both litigants and among the most-cited sources in the category. The AI Platform Citation Source Index 2026 puts Reddit as the single most-cited domain across generative engines at roughly 40% of multi-engine aggregate citation frequency, with Wikipedia second, appearing in 26–48% of ChatGPT top-10 answers.
The sources the engines depend on most are the ones with the strongest incentive and the deepest resources to contest the arrangement. Any strategy that assumes today's citation mix persists is assuming an outcome that is actively being litigated.
What is the regulatory layer doing?
Two instruments land very differently on a brand.
The EU AI Act's transparency obligations for general-purpose AI became fully enforceable on 2 August 2026. Those obligations fall primarily on model providers rather than on companies publishing content, and they reach a brand mainly through procurement: enterprise buyers increasingly ask suppliers how content was produced, whether it can be traced, and who signed it off.
Google's guidance is the one you control directly. Its published position is that using generative AI tools to generate many pages without adding value for users may violate its scaled content abuse policy, and it recommends adding information about how content was created. It sets no requirement to disclose AI authorship and no prohibition on AI-assisted content.
The pattern across both: the audit trail is becoming a commercial requirement before it is a legal one.
What follows for a brand?
- Assume structurally advantaged competitors in the citation set. Licensed outlets have a position in retrieval that no content programme matches. The realistic goal is being the source those outlets cite, not outcompeting them for the slot.
- Treat the citation mix as unstable for non-content reasons. A platform's share can move because of a contract or a ruling, not because your work changed. Programme design should survive that, which mostly means measuring rates over repeated runs rather than defending positions.
- Build provenance now, because procurement will ask. Claim, source, publisher, date, reviewer and production method, captured at the point of writing. Retrofitting is how a wrong figure acquires a citation.
- Do not treat licensing as a route you can buy into. These are arrangements with news organisations. There is no brand tier, and there is no indication one is coming.
- Read the price signal. Perplexity paying per citation is external evidence that a citation has value independent of clicks. Use it in the business case with the caveat attached.
- Keep your own record correctable. Where litigation or a contract removes a source that described you, the fallback is whatever remains — usually your own site and the directories.
The most common mistake here is treating publisher deals as a reason to disengage — "the big outlets have it sewn up". The data says otherwise. Engines cite between roughly 4 and 15 sources per answer depending on the surface, and more than half of tracked categories have no established owner. Licensed publishers take some slots. They do not take all of them.
Limits worth stating
- This is a moving picture. Every figure carries a date because deal counts, pool sizes and case counts all changed within the last year and will change again.
- Deal terms are mostly private. Public reporting covers who signed, rarely what was agreed, and almost never how it affects retrieval ranking.
- No causal evidence links licensing to citation share. It is a reasonable inference from the structure, not a measured effect, and this piece does not claim otherwise.
- Nothing here is legal advice. The regulatory position differs by jurisdiction and by how content is produced.
How Lifewood approaches this
Lifewood scopes AEO programmes on the assumption that the citation mix will move for reasons unrelated to the work, which changes two things in practice. Reporting is built on rates estimated from repeated runs rather than on positions, so a contractual or legal shift in the source set shows up as a change in the distribution rather than as an unexplained failure. And the third-party workstream is aimed at being the substrate that cited sources draw from, rather than at competing with licensed publishers for the same slot.
Provenance is captured at the point of writing rather than assembled afterwards: claim, source, publisher, date, named reviewer and production method, held together so a figure can be re-verified or retired rather than merely deleted. That record exists because enterprise procurement is beginning to ask for it, and because retrofitting sources to existing claims is how a wrong number acquires a citation.
Lifewood publishes these figures with their dates and their limits attached, including where the evidence is inference rather than measurement. See where AI answer engine citations go, how Perplexity picks sources and AEO services.
Sources and further reading
- LLM Pulse licensing tracker on OpenAI publisher deals, with eMarketer.
- LLM Pulse, Perplexity Publishers' Program and Comet Plus terms.
- Press Gazette, publisher AI lawsuits and licensing tracker.
- European Commission, EU AI Act transparency obligations effective 2 August 2026, via TechTimes.
- Google Search Central, guidance on AI-generated content.
- AI Platform Citation Source Index 2026, synthesis of six studies covering 680 million citations.
- Semrush with Kevin Indig, AI visibility is a topic-level game — category ownership figures.