AI chatbots use journalism for free: the media industry demands payment

Chatbots are siphoning off traffic while sources remain invisible

Millions of people increasingly get their news not from articles but from AI chatbot answers. According to a 2026 Reuters Institute survey covering more than 90,000 people worldwide, 10% use chatbots for news weekly, and among younger audiences that share is noticeably higher. This is reported by Deutsche Welle in a piece from the Global Media Forum (GMF) in Berlin.

A broken exchange: why publishers are sounding the alarm

In the past, search engines indexed publishers' websites and directed readers to them, bringing traffic in return. Funke Media Group publisher Julia Becker said at a GMF panel that this exchange has now broken down: AI companies rely on quality journalism to generate answers, but the original material ends up virtually invisible. DW Director-General Barbara Massing put the demand succinctly: 'AI systems must not simply use quality journalism. They must pay for it.'

What is already happening: lawsuits, coalitions and a German initiative

Pressure on the industry is coming from several directions. In October, German broadcasters ARD and RTL, together with the associations of digital and newspaper publishers (BDZV) and private media (VAUNET), called for stronger protection of media companies' copyright. In the US, The New York Times sued Brave Software, accusing it of using content without permission and bypassing the paywall. The European Court is examining whether AI answers themselves may infringe copyright. And in February, a media coalition — including the Guardian, BBC and Financial Times — demanded rules giving publishers control over AI systems copying their content.

Pay for answers or for training: lawyers differ on priorities

Professor Martin Senftleben, director of the Institute for Information Law at the University of Amsterdam, distinguishes two ways AI uses journalism: model training and generating answers in real time. Payments for training are legally more complicated; the more urgent problem is the use of material when answering a specific query. In his view, existing EU law already allows publishers and journalists to demand payment collectively, without waiting for new legislation: 'What are we waiting for? Germany is a large member state. This should be introduced in Germany.' He proposed a system of mandatory levies for every answer that draws on journalism — modelled on music royalty collection, with distribution through a collective management organisation. The rates of such a system could serve as a starting point in individual negotiations with AI companies.

The industry's position: readiness to negotiate and fear of regulation

Lutz Maché, senior manager for public policy at Google, said at the panel that the company is ready to negotiate with German media and discuss new AI products with them. Google is testing an 'AI Contribution' pilot with selected publishers — it is meant to reward content that matters for Gemini and AI Overviews answers, but it is an early experiment, not an open payment system. According to Google, its neighbouring rights agreements already cover more than 800 German publications, with payments being made. Microsoft, Anthropic and OpenAI have backed standardised signals allowing website owners to indicate whether their content may be used for training. However, lawyer Jörg Heidrich, AI commissioner at media company Heise, warned against new rules that ignore their effect on the tech sector: Europe is already lagging in the global AI race, and restrictions would hit startups, independent developers and research institutes that depend on access to data.