The Financial Times signs a deal with OpenAI to train AI models on the publisher's archived content and to let ChatGPT reply with short summaries of FT articles
Agreement comes as Microsoft-backed start-up seeks data from reliable sources to train latest artificial intelligence models
Financial TimesMadhumita Murgia
Context & Ripple Effects
The agreement extends OpenAI's publisher-partnership strategy after its AP archive licensing deal and subsequent arrangements with European news groups. It also arrives after the FT deployed its own Claude-powered Ask FT tool for a subset of professional subscribers.
The significance is not merely training access: the deal connects a publisher's archive to a consumer AI answer surface, making attribution, summary format, and referral pathways part of the commercial relationship.
First-order effects
OpenAI gains licensed access to the FT archive for model training and can return short FT article summaries in ChatGPT.
The FT creates a formal route for its reporting to appear in ChatGPT while retaining a defined partnership rather than leaving use of its archive solely to unlicensed web access.
Second-order effects
Other premium publishers face stronger pressure to decide whether to negotiate similar licenses, as Axel Springer's OpenAI partnership had already paired training access with ChatGPT news summaries.
AI answer products will need to make publisher sourcing and summary presentation commercially acceptable, because those choices affect whether publishers view them as distribution, substitution, or both.
Third-order effects
News archives are increasingly becoming licensable inputs to AI systems, shifting publisher AI strategy from standalone reader products toward negotiated participation in third-party answer interfaces.
If such deals broaden, access to authoritative, rights-cleared material could become a differentiator among AI providers, while publishers gain more leverage to set terms for use and attribution.
The trend: Generative-AI companies are moving from indiscriminate web-scale ingestion toward paid, publisher-negotiated access to high-quality content and answer distribution.
Under the terms of the deal, the FT will license its material to the ChatGPT maker to help develop generative AI technology that can create text, images and code indistinguishable from human creation https://www.ft.com/... [image]
Big publishers will fight their corner. They arguably should. (I would, in their shoes.) Serves their self-interest, and they need to make money to do journalism. But unless proven otherwise, I'd hold off on seeing any deals as serving a wider, let alone diverse, public interest
“The agreement also allows ChatGPT to respond to questions with short summaries from FT articles” — the FT is dismantling its own paywall via generative AI https://www.ft.com/...
More and more of these licencing agreements popping up as AI companies seek to avoid expensive lawsuits The Financial Times and OpenAI strike content licensing deal https://t.co/... via @ft
Winner-takes-most strikes again. As FT signs licencing deal with Open AI, joining AP, Axel Springer, Le Monde, and Prisa, I think back to how our survey of 300+ digital leaders in news found 35% expected most $$$ from deals would go to big media companies https://reutersinstitute…