Facing off against Google, Qwant is experimenting with an AI that pays news publishers.

Between Google's AI-powered direct responses and chatbots like ChatGPT, usage patterns are evolving, but this also raises a question: that of compensation for the content used to train or feed these tools.

In this context, the French search engine Qwant has announced the launch of an experiment with several major media outlets and press publishers, with the goal of testing a generative AI feature integrated into the search engine… while implementing a revenue-sharing model with rights holders.

“Flash Response,” Qwant’s AI feature

A concrete example of a "Flash Answer" – Source: Qwant

At the heart of this experiment is a feature called "Flash Answer". Specifically, an answer appears at the top of the search results in the form of a box with an AI-generated summary, as well as contextual information.

The principle is very reminiscent of Google's AI Overviews, which synthesize information from the web to provide a quick answer. However, the difference lies in the method adopted by Qwant. To produce these summaries, the French search engine relies on the Mistral language model, which analyzes several partner sources before generating the answer, in order to cross-reference information and improve reliability. Around twenty media outlets are already participating in this experiment, including Le Figaro, Les Échos, Ouest-France, Le Parisien, Paris Match, RMC BFM, and France Médias Monde, as well as the regional press group Ebra. The project began in January and is scheduled to last nine months, in order to observe usage and measure the value created by this type of functionality. A 50/50 business model is also being considered. The other aspect of this initiative directly concerns content remuneration. Unlike some digital players accused of capturing the value generated by journalistic content, starting with Google, Qwant offers a revenue-sharing model. The advertisements displayed in the "Flash Answer" space generate revenue that is then divided equally between Qwant and its partner publishers. Media outlets also benefit from transparency regarding the use of their content in AI-generated answers. According to Olivier Abécassis, CEO of Qwant, this approach aims to build "a constructive model for the ecosystem" and to develop more transparent innovation. The company also emphasizes its technological sovereignty, relying on European technologies. For organizations representing publishers, this approach could open a new path. Marc Feuillée, president of the Alliance of General Information Press, believes that this experiment provides, for the first time, visibility into the use of content in AI-generated responses. If the results are deemed convincing at the end of the testing phase, the initiative could help redefine the relationship between search engines, artificial intelligence, and the media.