Since mid-2023, news organisations, academic publishers and online platforms have announced a steady stream of partnerships with AI companies. Two of the first were Shutterstock's and the Associated Press's agreements with OpenAI in July 2023. By October 2026, Google said more than 400 publications were taking part in its News AI pilot programme. Most of these agreements give an AI company access to content, whether archive material for training, current reporting for AI answers, or both. Many also include attribution and links back to the publisher, access to the AI company's technology, or both.
This article sets out a dated timeline of publicly announced partnerships, drawn from company announcements and reputable press coverage, then describes the patterns that run through them and what content owners can prepare. It records what each partnership was announced to cover; it does not assess any agreement.
Key takeaways
- Publicly announced partnerships between publishers and AI companies began in 2023 and grew to hundreds of participating publications by late 2026.
- Early deals centred on archive access for model training, while later ones increasingly cover current content shown in AI answers, with attribution and links.
- Many agreements also give publishers access to AI technology, from tools for staff to help building new products.
- Newer arrangements include marketplaces and programmes that bring many publishers together, rather than one agreement at a time.
- Partnerships reach beyond news to online platforms such as Reddit, image libraries such as Shutterstock, and academic publishers such as Taylor & Francis and Wiley.
How to read this timeline
- Dates are given to the month in which each partnership was announced or first reported.
- Entries describe what was announced: what the companies, or reputable press reports, said each partnership covers. Other details are often not made public.
- The list is selective. It includes partnerships confirmed by the companies involved or reported by reputable outlets. Press Gazette's running tracker, the source for many entries here, lists more.
2023: the first agreements
| Month | Parties | What was announced |
|---|---|---|
| July 2023 | Shutterstock and OpenAI | A six-year agreement providing images, video, music and metadata as training data, with Shutterstock getting priority access to new OpenAI technology |
| July 2023 | Associated Press and OpenAI | OpenAI gained access to part of AP's text archive, with stories going back to 1985, and AP gained access to OpenAI's technology and product expertise |
| December 2023 | Axel Springer and OpenAI | ChatGPT to provide summaries of content from brands including Politico, Business Insider, Bild and Welt, including paywalled material, with attribution and links. The content could also be used for training |
2024: news groups, platforms and academic publishers
| Month | Parties | What was announced |
|---|---|---|
| February 2024 | Reddit and Google | Google gained access to Reddit's Data API, which it described as delivering "real-time, structured, unique content", to display and train on. Reddit would use Google's Vertex AI for its own AI features |
| March 2024 | Le Monde, Prisa Media and OpenAI | ChatGPT to offer summaries with attribution and links to content from Le Monde and Prisa titles including El País; content also usable for training |
| April 2024 | Financial Times and OpenAI | The first major UK news brand to announce a deal with OpenAI, covering both current news and archive material |
| May 2024 | Informa (Taylor & Francis) and Microsoft | Informa, parent company of the academic publisher Taylor & Francis, disclosed a non-exclusive agreement giving Microsoft access to its Advanced Learning content and data, alongside work on AI tools |
| May 2024 | News Corp and OpenAI | Content from titles in the US, UK and Australia, including The Wall Street Journal, The Times and The Australian, for use in OpenAI's models and products |
| May 2024 | The Atlantic, Vox Media and OpenAI | Content to surface in ChatGPT with attribution and links, alongside access to OpenAI technology for the publishers' own products |
| June 2024 | Time and OpenAI | Access to Time's 101-year archive and current reporting for cited, linked answers, with Time gaining access to OpenAI technology |
| July 2024 | Perplexity Publishers' Program | First members included Time, Der Spiegel, Fortune, Entrepreneur, The Texas Tribune and Automattic, with access to Perplexity's technology and analytics |
| August 2024 | Condé Nast and OpenAI | A multi-year partnership to display content from titles such as Vogue, The New Yorker and Wired in ChatGPT and its search features, with attribution and links |
| October 2024 | Hearst and OpenAI | Content from more than 20 US magazines and 40 newspapers in ChatGPT, with citations and links |
| October 2024 | Reuters and Meta | Meta's first news-publisher AI deal: real-time Reuters content used to answer news questions in Meta AI, with summaries and links |
2025: real-time answers and more AI companies
| Month | Parties | What was announced |
|---|---|---|
| January 2025 | Axios and OpenAI | A three-year partnership: Axios content in ChatGPT with attribution and links, and OpenAI technology to support Axios opening local newsrooms in four new cities |
| January 2025 | Associated Press and Google | A feed of real-time information from AP to improve results in the Gemini app |
| January 2025 | AFP and Mistral | A multi-year agreement giving Mistral's Le Chat access to AFP's text output in six languages |
| February 2025 | Schibsted and OpenAI; The Guardian and OpenAI | Real-time articles and attributed summaries in ChatGPT, with access to OpenAI technology |
| April 2025 | The Washington Post and OpenAI | Summaries, quotes and links to Post reporting in ChatGPT search, with clear attribution |
| May 2025 | Le Monde and Perplexity | Le Monde content in Perplexity's answers, with links, but not used for training, plus access to Perplexity's technology |
| May 2025 | The New York Times and Amazon | Real-time summaries and short excerpts in Amazon products such as Alexa, plus use in training Amazon's models |
| June 2025 | Snowflake with AP, USA Today and other publishers | Cortex Knowledge Extensions launched, letting companies bring publishers' current news into their own AI applications and agents |
| July 2025 | Wiley and Anthropic | A pilot using the open Model Context Protocol to bring Wiley's peer-reviewed content into Claude, starting with Claude for Education, with author attribution and citations |
| July 2025 | Condé Nast, Hearst and Amazon | Multi-year agreements covering content for Amazon's Rufus shopping assistant |
| July 2025 | Gannett (USA Today Network) and Perplexity | Content from USA Today and more than 200 local titles in Perplexity's answers and Comet browser, plus access to Perplexity's technology |
| November 2025 | People Inc and Microsoft | People Inc said it was joining Microsoft's planned Publisher Content Marketplace, with Copilot as the first AI product to use it |
| December 2025 | Meta with CNN, Fox News, Le Monde Group, People Inc, USA Today and others | Real-time content for Meta AI, with links to the publishers |
| December 2025 | Google News AI pilot | Launched with publishers including Der Spiegel, El País, Folha, The Guardian, The Times of India and The Washington Post, testing AI article overviews and audio briefings in Google News |
2026: marketplaces and scale
| Month | Parties | What was announced |
|---|---|---|
| February 2026 | Microsoft Publisher Content Marketplace | Microsoft introduced the marketplace, in pilot, as a place where AI builders can find and use publisher content for specific uses, with usage reporting for publishers. It was co-designed with publishers including AP, Condé Nast, Hearst Magazines, People Inc and Vox Media |
| March 2026 | Reach and Amazon | Reach's first AI deal: its content used in Amazon's Alexa assistant and Nova AI model |
| March 2026 | News Corp and Meta | A multi-year agreement covering US and UK content: archive material for training, and new reporting retrieved for users of Meta's AI products |
| May 2026 | Folha de S.Paulo, UOL and OpenAI | Real-time Portuguese-language news in ChatGPT, with summaries linking to the original reports; OpenAI's first agreement with Brazilian media companies |
| July 2026 | Nine and Microsoft | Real-time reporting from titles including The Sydney Morning Herald and The Age to ground Copilot answers, with summaries that link back to Nine's sites |
| September 2026 | Village Media and OpenAI | OpenAI's first news partnership in Canada: technical help to build Open Door, an AI tool for finding local community services, with Village Media's reporting cited in ChatGPT |
| September 2026 | Times Group and Indian Express Group with OpenAI | Summaries and excerpts with links in ChatGPT from titles including The Times of India and The Indian Express. Both cover current and archive content, and the Indian Express agreement spans seven languages |
| October 2026 | Google News AI pilot expansion | Google said more than 400 publications were taking part, including Axios, CNN, the Financial Times and Forbes. Features such as article overviews and audio briefings come "with clear attribution and direct links" |
Patterns across the deals
A handful of components recur across the timeline, and most agreements combine several.
Archive access
The earliest deals were about the back catalogue. AP's archive went back to 1985, and Time opened its 101-year archive. Later agreements, such as News Corp's with Meta and the Indian publishers' with OpenAI, still name archives explicitly. For training, a large, edited, consistently structured archive is what a publisher brings. Our guide to how AI models use data explains how developers weigh different sources in training mixes.
Real-time content for answers
From 2024 onwards, more deals focused on current reporting appearing in AI products: ChatGPT search, Meta AI, the Gemini app, Copilot, Alexa and Perplexity. Here the AI system retrieves content when a user asks, rather than learning from it in advance. Some agreements keep the two uses apart: Le Monde's agreement with Perplexity covers answers but not training. Snowflake applies the same retrieval idea inside companies' own AI tools.
Attribution and links
Many announcements describe content appearing with attribution and links, including Axel Springer, Le Monde and Prisa, Hearst, Reuters, Nine and Google's News AI pilot. For publishers, links are how an AI answer can send readers back to the original site. Wiley's partnership with Anthropic applies the same idea to research, with author attribution so that insights can be traced back to their sources.
Technology collaboration
Access to AI tools often travels with content. AP gained access to OpenAI's technology and product expertise. The Atlantic and Vox Media gained access to OpenAI technology for their own products. Axios and Village Media are building new local services with OpenAI's help. Several 2025 and 2026 agreements, including Gannett's with Perplexity and Folha and UOL's with OpenAI, include access to AI products for staff.
Marketplaces and programmes
Later arrangements bring many publishers into one structure, rather than one agreement per publisher, as Perplexity, Snowflake and Microsoft's marketplace do. Microsoft said its marketplace is "intended to scale to avoid the challenge of pairwise agreements between every publisher and every AI builder and agent". Google's pilot does the same at scale, with more than 400 publications.
Beyond news
Google described the content it receives from Reddit as "real-time, structured, unique": user-generated discussion is part of this market too. Shutterstock's agreement covered images, video and music. The agreements announced by Informa (Taylor & Francis) and Wiley extend the same pattern to scholarly content.
What the timeline suggests for content owners
The Reuters Institute's 2026 trends report observes that "so far most of the big deals have been with upmarket newspapers and news agencies". In its survey of 280 senior news leaders in 51 countries and territories, carried out in late 2025, putting more effort into navigating distribution through AI platforms scored a net +61, against −25 for old-style Google search optimisation.
Whatever an organisation decides, the deals above point to practical groundwork that makes any conversation easier:
- Know what the archive contains. Coverage by date, topic, language and format is basic information any counterpart is likely to need. See what makes content valuable to AI developers.
- Separate archive and current content. The timeline shows these are often treated differently: archives for training, current output for real-time answers.
- Make content machine-deliverable. Several real-time deals involve structured delivery, such as Reddit's Data API and AP's real-time feed for Gemini. The Reuters Institute notes that AI-tailored formats require media companies to move "away from authoring 'articles' towards more flexible atomic objects".
- Keep canonical URLs and metadata. Attribution and linking only work if each item has a stable address, a date and a byline.
- Set a deliberate crawler policy. How a site treats AI crawlers is a separate decision from any partnership. Our practical guide to AI crawlers and robots.txt covers the options.
How Druid Learning helps
Druid Learning is an AI archival processing platform, based in Dublin and New York, that turns content archives into AI-ready datasets. Through Data Partnerships, Druid prepares an archive, raw or structured, and connects content owners with AI companies. Its Archive Optimization product turns archives in any format into an indexed, labelled, AI-ready dataset. Druid has processed 10M+ assets and handles content in 140+ languages. Its work includes 2M+ assets processed for NPR and 50,000 curriculum-mapped resources for the Irish educational publisher CJ Fallon.
Frequently asked questions
What was the first major publisher deal with an AI company?
Two of the earliest were announced in July 2023. Shutterstock agreed to provide OpenAI with images, video, music and metadata as training data. Days later, the Associated Press and OpenAI agreed that OpenAI would access part of AP's text archive, going back to 1985, while AP gained access to OpenAI's technology and product expertise. Axel Springer followed in December 2023, with summaries of its journalism in ChatGPT.
Which AI companies have content partnerships with publishers?
Most major AI developers have announced at least one. OpenAI appears most often in this timeline, from AP and Axel Springer in 2023 to Folha, UOL and Indian publishers in 2026. Google, Microsoft, Meta, Amazon, Perplexity and Mistral have their own agreements, Anthropic has a research partnership with Wiley, and Snowflake offers publisher content inside companies' own AI tools.
Do these partnerships only cover AI training?
No. Early deals such as AP's focused on archive access for training, but many later agreements cover showing current content in AI products, such as summaries and links in ChatGPT, Meta AI, Gemini or Copilot. Some, such as Le Monde's agreement with Perplexity, cover answers but not training. Others add access to AI technology, and programmes such as Google's News AI pilot bring hundreds of publications into one arrangement.
Are academic publishers part of the AI content market?
Yes. In May 2024 Informa, the parent company of Taylor & Francis, disclosed a non-exclusive agreement giving Microsoft access to its Advanced Learning content and data. In July 2025 Wiley announced a partnership with Anthropic that uses the Model Context Protocol to bring peer-reviewed content into Claude, starting with Claude for Education, with author attribution so that readers can trace insights back to the original sources.
Sources
- News generative AI deals revealed — Press Gazette
- OpenAI and AP announce news content deal (July 2023) — Associated Press (via KSAT)
- An expanded partnership with Reddit — Google
- Snowflake retools data cloud to fuel AI agents, expand integrations — CIO Dive
- Wiley Partners with Anthropic on AI Integration for Scholarly Research — Publishers Weekly
- Building Toward a Sustainable Content Economy for the Agentic Web — Microsoft Advertising
- 400 publications are now partnering with us on our News AI pilot program — Google
- Journalism, media, and technology trends and predictions 2026 — Reuters Institute