OpenAI Rules Out Ad Revenue Sharing for Publishers as ChatGPT Ads Launch

Digiday has characterized this as a "bitter pill" for publishers: OpenAI generates additional revenue from ads served alongside content originally created by news organizations, but no share of that money flows back to the creators .

What OpenAI Offers Publishers Instead

Rather than an ongoing ad revenue split, OpenAI structures its publisher relationships around two alternative forms of value.

Upfront content licensing deals are the primary vehicle. The company has signed multi-year, fixed-fee agreements with dozens of major publishers for access to archives and training data. Dotdash Meredith's deal is reportedly worth approximately $16 million per year, and partners include Vox Media, The Atlantic, and Axel Springer, among others . These are one-time or annual lump-sum payments — not ongoing revenue shares tied to actual usage.

Preferential distribution is the second component. A leaked "Preferred Publisher Program" reportedly grants select partners priority placement and "richer brand expression" in ChatGPT responses, effectively trading visibility for content access .

Notably, Shetty has publicly stated that he does not see traffic as the "core value" of appearing in ChatGPT search results . This represents a fundamental break from the traditional search engine model, where Google drove billions of visits to publisher sites as its primary exchange of value.

Where Competitors Stand: A Tale of Three Models

The revenue-sharing landscape has fragmented into three distinct approaches, with OpenAI on one end and its smaller competitors on the other.

Prorata AI: The 50/50 Model

Prorata AI represents the most aggressive revenue-sharing model in the market. The startup operates Gist.ai, an AI search engine built on a foundational promise to split 50% of all revenue with publisher partners on a recurring basis . Prorata's system uses proprietary attribution algorithms to determine how much each publisher's content contributed to a given AI response and distributes payments proportionally .

The company has attracted notable publishing partners including the Boston Globe, Future, Vox Media, Der Spiegel, The Atlantic, the Financial Times, and the Texas Tribune, along with backing from the News/Media Alliance . ProRata pitches itself as a "neutral platform" built to ensure equitable compensation from the start .

Perplexity: The Subscription Pool Model

Perplexity launched its Comet Plus revenue-sharing program in 2025, setting aside a $42.5 million pool to compensate publishers when their content is cited in answers . The model works differently from Prorata's: revenue from Perplexity's subscription tiers (Pro, Max, and Comet Plus) is pooled, and 80% of it is distributed to participating publishers based on three categories — direct visits, crawler traffic, and AI agent usage .

Perplexity initially also shared ad revenue when it launched advertising in late 2024 but has since removed advertising from its platform entirely . Despite signing a large number of publisher partners, Perplexity has faced trust challenges. Some publishers told Digiday they "couldn't get into" the program, and payouts from Perplexity have been described as "a fraction" of what OpenAI offered in upfront licensing deals .

The Core Differences at a Glance

The three models break down into fundamentally different compensation philosophies:

• OpenAI pays upfront licensing fees with no ongoing ad revenue share, offering fixed multi-year contracts that compensate for data access regardless of how often content is actually used.
• Prorata AI distributes 50% of all revenue to publishers proportionally based on content usage, creating an ongoing financial incentive tied directly to how much value a publisher's work provides.
• Perplexity pools subscription revenue and distributes it based on a combination of direct traffic, citations, and agent usage, with payouts that have proven smaller than OpenAI's upfront deals in many cases.

Sulzberger's 'Brazen Theft' Accusation and the Legal Front

The debate over revenue sharing is unfolding against a backdrop of escalating legal and rhetorical warfare. On June 1, 2026 — the day before OpenAI's Shetty confirmed no ad revenue sharing — New York Times publisher and chairman A.G. Sulzberger delivered the opening keynote at the same WAN-IFRA Congress in Marseille .

Sulzberger accused AI companies of "brazen theft of intellectual property" committed on an "unprecedented scale" . He described tech companies as "strip-mining" news websites without permission or compensation and warned that their repackaging of journalistic content amounted to dealing in "stolen goods" . According to Sulzberger, the New York Times was the main source of proprietary data in one widely-used AI training dataset .

The rhetoric is backed by real legal costs. The New York Times has spent more than $20 million on lawsuits against OpenAI, Microsoft, and Perplexity since filing its first case in December 2023 . The Times sued Perplexity separately in December 2025 for copying journalism "without permission or compensation" through retrieval-augmented generation methods . Other publishers, including CNN and Alden Global Capital, have filed or joined their own suits .

Sulzberger also revealed that the New York Times spent more than $2 billion in 2025 alone producing nearly half a million pieces of journalism — a figure meant to underscore the scale of the investment that AI companies are accused of appropriating .

The Publisher's Dilemma: License, Sue, or Bet on Revenue Sharing

For news publishers, the current landscape presents an uncomfortable set of choices. Accepting an OpenAI-style upfront licensing deal provides guaranteed cash but forecloses participation in any future revenue growth — particularly as the company targets $25 billion in annual ad revenue . Holding out and betting on revenue-sharing newcomers like Prorata potentially aligns incentives but means waiting for a smaller platform to scale. Suing, as the New York Times has done, offers a chance to set legal precedent but consumes enormous resources with uncertain outcomes.

The fundamental tension, as Shetty's comment about traffic not being the "core value" reveals, is that AI companies and publishers increasingly see the exchange differently. Publishers want ongoing participation in the value their content creates. OpenAI sees the transaction as complete once a licensing check clears.

The European Union's AI Act already imposes transparency requirements on training data usage, and the UK has been active through its AI Safety Institute and Intellectual Property Office consultations on AI and copyright. These regulatory actions signal that governments are moving to give publishers more leverage — but at the speed of legislation, not technology.